The communications technology journal since 1924
90
1924-2
Delivering content
with LTE Broadcast 4
Nine decades of innovation
11, 19 & 47
Non-line-of-sight microwave
backhaul for small cells 12
Software-defined networking: the
service provider perspective 20
HSPA evolution: for future mobilebroadband needs 26
Next generation video
compression 33
Next generation
OSS/BSS architecture 38
Carrier Wi-Fi:
the next generation 48
014
Editorial
Celebrating 90 years
of technology insights
“The object of this magazine is to
spread information concerning
the work and activities of this and
associated enterprises, and to
furnish a connecting link between
these latter and the head firm.”
These lines are taken from the
introduction to The L. M. Ericsson
Review – Tidskrift för Allmänna
Telefonaktiebolaget L. M. Ericsson –
when the first issue of the new journal
was published in 1924.
In his historical article to celebrate
this journal’s 50th anniversary (1974),
Sigvard Eklund, former editor of
Ericsson Review (1943-1972), wrote the
following words:
“Apart from an article on ‘the development and present size of the LM
Ericsson Group,’ there was a 10-page
description of the company’s automatic 500-line selector system, illustrated by a few photographs of the
recently opened automatic exchange
in Rotterdam, one of the first major
exchange equipments to be delivered
up to that time.”
In the 40 years since then – and the
90 years since this journal first started
promoting technology – the world we
live in has been transformed by technology to such a degree that I sometimes find it difficult to recognize the
old one. I would like nothing more
than to be able to give you a glimpse
of what we will be writing about 90
years from now… what will the 22nd
century bring? What generation of
technology will have been reached by
then, what business models will we
use, how will we pay for things, and
what sort of devices will we connect
with? These are just some of my questions, and I wonder even if they will
be relevant; maybe we won’t even use
devices, as connectivity will simply
exist in everything. Even if I don’t have
an open window on the next century, the research and development we
carry out at Ericsson today is aimed at
the next generation, which promises
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
to continue along the current path
of evolution: to be data driven, video
heavy and influenced by the gaming
world.
The articles in this edition address
a wide range of telecommunication
issues, but they all have one thing in
common and that is performance.
Getting data through the network fast
and efficiently so that the best user
experience can be delivered to subscribers is a recurring theme no matter what part of network architecture
is being discussed. From future OSS/
BSS architecture to integrated Wi-Fi
and to packets stuffed with data, the
message is clear… the faster the network can serve one subscriber, the
faster it can move on to the next.
One thing I am convinced about,
however, is convergence. And not
just in terms of fixed and mobile, but
everywhere. Industries and technologies are merging. The lines between
TV, the internet, and telecommunication will not exist for much longer.
Education, work and family life are
all coming together and the key is
individualism.
With a connection, every individual on the planet has the potential to
take control over their life. The traditional models of work and education
are being challenged. Connectivity
is providing individuals with more
choices, greater flexibility and the ability to mix things up in a way that suits
them, their budget, their lifestyle and
their goals. We are not there yet, and
there are many pieces that need to
be in place, but right now we are laying the foundation for the Networked
Society. Mobile subscriptions are set to
rise to 9.3 billion and mobile data traffic to grow by 45 percent (CAGR) by
2019. The opportunities are becoming
available for more people, and connectivity is becoming a way of life.
This edition is a celebration of 90
years of technology innovation. I hope
you enjoy it.
The most frequent
users interact with their
smartphone more than
150 times a day, or
an average of every
seven minutes during
the daytime.*
*Ericsson Mobility Report, November 2013
Ulf Ewaldsson
Chief Technology Officer
Head of Group Function
Technology at Ericsson
The communicat ons techno ogy journal since 1924
CONTENTS
90
1924-20
90TH ANNIVERSARY 2014
4
a collection of articles from 2013
Delivering content
with LTE Broadcast 4
Nine decades of innovation
11 19 & 47
Non line of sight microwave
backhaul for small cells 12
Software defined networking the
service provider perspective 20
HSPA evolution for future mobile
broadband needs 26
Next generation video
compression 33
Next generation
OSS/BSS architecture 38
Carrier Wi Fi
the next generation 48
4 Delivering content with LTE Broadcast
The data volume in mobile networks is booming – mostly due to the success of smartphones
and tablets. LTE Broadcast is one way of providing new and existing services in areas that can at
times be device dense, such as stadiums and crowded city centers. Built on LTE technology, LTE
Broadcast extends the LTE/EPC with an efficient point-to-multipoint distribution feature that can
serve many devices with the same content at the same time.
This article was originally published on February 11, 2013.
To bring you the best of Ericsson’s
research world, our employees have
been writing articles for Ericsson
Review – our communications
technology journal – since 1924.
Today, Ericsson Review articles have
a two-to-five year perspective and
our objective is to provide you with
up-to-date insights on how things are
shaping up for the Networked Society.
Non-line-of-sight microwave backhaul
for small cells
Address :
Ericsson
SE-164 83 Stockholm, Sweden
Phone: +46 8 719 00 00
The evolution to denser radio-access networks with small cells in cluttered urban environments
has introduced new challenges for microwave backhaul. A direct line of sight does not always
exist between nodes, and this creates a need for near- and non-line-of-sight (NLOS) microwave
backhaul. This article was originally published on February 22, 2013.
Publishing:
Ericsson Review articles and
additional material are pub ished
on www ericsson.com/review. Use
the RSS feed to stay informed of the
latest updates.
Articles are also
available on the Ericsson
Technology Insights app
for Android and Apple
tablets. The ink for your device is on
the Ericsson Review website:www.
ericsson.com/review. If you are
viewing this digitally, you can:
download from Google Play or
download from the App Store
Publisher: U f Ewaldsson
Editorial board:
Håkan Andersson, Hans Antvik,
Ulrika Bergström, Joakim Cerwall,
Deirdre P. Doyle, Dan Fahrman,
Anita Frisell, Jonas Högberg,
U f Jönsson, Magnus Karlsson,
Cenk Kirbas, Sara Kullman,
Kristin Lindqvist, Börje Lundwall,
Hans Mickelsson, U f Olsson,
Patrik Regårdh, Patrik Roséen and
Gunnar Thrysin
Editor:
Deirdre P. Doyle
deirdre.doyle@jgcommunication se
Chief subeditor:
Birgitte van den Muyzenberg
Contributors:
John Ambrose, Håkan Andersson,
Paul Eade, Ian Nicholson,
Gunnar Thrysin and Peter Öhman
Art director and layout:
Jessica Wiklund and Carola Pilarz
Illustrations:
Claes-Göran Andersson
Printer: Edita Bobergs, Stockholm
ISSN: 0014-0171
Volume: 91, 2014
11, 19 & 47 Nine decades of innovation
Automatic exchanges to smart networks.
12 Software-defined networking: the service provider
perspective
20 An architecture based on software-defined networking (SDN) techniques gives operators greater
freedom to balance operational and business parameters, such as network resilience, service
performance and QoE against opex and capex. With its beginnings in data-center technology,
SDN has developed to the point where it can offer significant opportunities to service providers.
This article was originally published on February 21, 2013.
26 HSPA evolution for future mobile-broadband needs
As HSPA evolution continues to address the needs of changing user behavior, new techniques
develop and become standardized. This article covers some of the more interesting techniques
and concepts under study that will provide network operators with the flexibility, capacity and
coverage needed to carry voice and data into the future, ensuring HSPA evolution and good user
experience. This article was originally published on August 28, 2013.
33 Next generation video compression
Requiring only half the bitrate of its predecessor, the new standard – HEVC or H.265 – will
significantly reduce the need for bandwidth and expensive, limited spectrum. HEVC (H.265) will
enable the launch of new video services and in particular ultra-HD television (UHDTV). This article
was originally published on April 24, 2013.
38 Next generation OSS/BSS architecture
When two large companies merge, it often takes a while – years in some cases – before
processes get redesigned to span all departments, and the new organization settles into a lean
and profitable machine. And the same is true of OSS/BSS. These systems have been designed
for two different purposes: to keep the network operational and to keep it profitable. But today’s
demanding networks need the functions of both of these systems to work together, and to work
across the varying life cycles of products and services. This article was originally published on
November 25, 2013.
48 Carrier Wi-Fi: the next generation
Putting the network in control over whether or not a device should switch to and from Wi-Fi, and
when it should switch, will make it easier for operators to provide a harmonized mobile broadband
experience and optimize resource utilization in heterogeneous networks. This article was
originally published on December 20, 2013.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Capture your audience
4
Delivering content
with LTE Broadcast
Ericsson has demonstrated LTE Broadcast with evolved Multimedia Broadcast
Multicast Services at a number of international trade shows. These demos have shown
the solution’s potential to create new business models for telcos and ensure consistent
QoS, even in very densely populated places like sports venues.
T HOR S T E N L OH M A R , M IC H A E L S L S S I NGA R , V E R A K E N E H A N A N D S T IG P U U S T I N E N
The solution is built on LTE
technology, extending the LTE/
EPC with an efficient point-tomultipoint distribution feature
that can serve many eMBMScapable LTE devices with the
same content at the same time.
It can be used to boost capacity
for live and on-demand content so
that well-liked websites, breaking
news or popular on-demand video
clips can be broadcast – offloading the network and providing
users with a superior experience.
Single-frequency network (SFN) technology is used to distribute broadcast
streams into well-defined areas – where
all contributing cells send the same data
during exactly the same radio time slots.
The size of the coverage area of an LTE
SFN can vary greatly, from just a few
cells serving a stadium, to many cells
delivering content to an entire country. eMBMS-enabled devices can select
BOX A the broadcast streams within the SFN
that are of interest. In this way, devices
download only relevant data – not everything within the area to then just throw
unwanted data away. This ensures that
devices work in a battery-efficient way.
respondents stated they would watch
more TV if the content was provided
on their mobile device, and 61 percent
said they would switch operator to gain
access to mobile-TV services. The majority of respondents said content they
would find interesting to watch while
on the move includes local news and
weather information, movies, national
news, sitcoms and sports.
To meet this growing demand for
mobile TV, operators are rapidly updating their offerings, continuously adding new services and content to live
and on-demand streams – increasing
the volume of information transported. Naturally, this causes network utilization to rise, requiring more efficient
ways to deliver content, while network
dimensioning becomes all the more crucial, and new business models are needed to maintain ARPU.
Given the direction in which the
industry is clearly moving, Ericsson has
developed an end-to-end LTE Broadcast
Business incentives
The coextending evolution of mobile
technologies and devices has made it
possible for people to consume video
using handheld equipment without
compromising their experience. Based
on an Ericsson ConsumerLab study1, the
most recent Ericsson Mobility Report2,
states that video is the biggest contributor to mobile-traffic volumes, accounting for more than 50 percent. And the
growth of traffic is expected to continue, increasing 12-fold by 2018.
According to another study, carried
out by Mobile Content Venture3, more
than half of US consumers would consider viewing programs on their smartphones and tablets – 68 percent of
Terms and abbreviations
AL-FEC
Application Layer FEC
API
application program interface
ARPU
average revenue per user
BLER
Block Error Rate
BM-SC
Broadcast Multicast Service Center
CDN
content distribution network
eMBMS
evolved MBMS
eNB eNodeB
EPC
Evolved Packet Core
EPS
Evolved Packet System
FDD
frequency division duplex
FEC
forward error correction
FIFA
Fédération Internationale de Football Association
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
FLUTE
file delivery over unidirectional transport
HEVC
High Efficiency Video Coding
IMB
integrated mobile broadcast
ISD
inter-site distance
ISI
inter-symbol interference
M2Mmachine-to-machine
MBMS
Multimedia Broadcast Multicast Service
MBMS-GW MBMS-gateway
MBSFN
Multimedia Broadcast over an SFN
MCE
Multicell Coordination Entity
MME
Mobility Management Entity
MPEG
Moving Picture Experts Group
MPEG-
DASH
NBC
OFDM
PGW
SDK
SFN
SGW
SNR
TDD
UDP
UE
MPEG-Dynamic Adaptive
Streaming over HTTP
National Broadcasting Company
orthogonal frequency division multiplexing packet data network gateway
software development kit
single-frequency network
service gateway
signal-to-noise ratio
time division duplex
User Datagram Protocol
user equipment
5
FIGURE 1 Broadcast versus unicast
Y%, Y%
Y%, Y%
X%
Y%,Y%
Unicast
Broadcast
Table 1: Broadcast versus unicast
Broadcast
One data channel
per content
Limited data channels,
unlimited number of users
Resource allocation viewer
independent
Unicast
One data channel
per user
Unlimited channels, limited
number of users
Resources allocated when needed
solution. The concept has been built on
eMBMS technology and based on a set
of use cases that can be divided into two
main categories
delivery of live premium content; and
unicast off-loading (for example, local
device caching).
Premium content
Despite the diversity of available content and an obvious shift by subscribers
towards on-demand viewing, watching
certain events and programs live continues to appeal to large audiences.
London 2012 is a good example of
an event that enjoyed widespread liveviewing appeal. Ratings place the NBC
coverage of the games as some of the
most watched TV in US history; almost
half of the online video streams were
delivered to tablets or smartphones,
and revenue expectations were far surpassed. Some use cases for premium
content follow.
Regional and local
This use case covers regional and local
interest events, such as concerts, sports
fixtures or breaking news. Such as the
Super Bowl, FIFA World Cup matches,
as well as elections and royal weddings.
Given suitable content security and digital-rights handling, this use case can
be enhanced to allow users to store and
replay the event on-demand from their
device for a certain period of time.
this technology also supports file delivery. Exploiting this and the caching
capability available in both mobile and
fixed devices creates new possibilities
for a range of use cases.
Venue casting
Popular content
This use case covers specific locations
such as shopping malls, museums, airports, university campuses and amusement parks. In this case, the operating
enterprise may wish to broadcast content to users, which can vary from
breaking news of national interest
to very specific information such as
special offers available at the mall,
­additional information about the main
artist of an art exhibition, or departures
and a
­ rrivals information at the airport.
For all of these premium-content use
cases, operators can deliver services on
a nationwide basis as well as ­locally.
The duration of a broadcast and the
size of the geographical area where it
is available can be managed dynamically, depending on the nature and relevance of the content. By using unicast
for blended services, broadcast services
can be complemented with interactivity – opening up new ways to generate
revenue from content.
At a soccer match, for example, these
value-added services could include
­video streams carrying footage from
additional camera angles, diverse audio
coverage and live results of related
matches taking place at the same time
in other stadiums.
Operators can choose to deliver popular TV and video clips to the local cache
of a user’s device at their convenience.
Based on content popularity and busyhour-traffic distribution, operators can
deliver content when network load is
low. Content shared on popular video streaming sites, as well as the content provided by national and cable
TV channels can all be pre-loaded to
mobile devices through broadcast – significantly reducing the overall network
capacity required to deliver frequentlyconsumed video streams.
Unicast off-loading
MBMSs are traditionally associated with
the delivery of live, linear TV, although
News
Daily clips and subscription content
such as a magazine can be pre-delivered
to the cache of a subscriber’s preferred
device for that content.
Software upgrades
Upgrades to application software and
operating systems are usually released
over the network to large numbers of
subscribers at the same time. This traditional way of performing an upgrade
can be a burden on the network. By
using LTE Broadcast instead, upgrades
can be distributed as packages to a
multitude of devices at little expense
in terms of required resources – an
approach that is particularly advantageous if the broadcast can be delivered
during off-peak hours.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Capture your audience
6
FIGURE 2 the radio-channel quality and overall
traffic volumes within the cell.
Broadcast is implemented as an
extension to the existing EPS architecture (see Figure 7 and Box B). Ericsson’s
LTE Broadcast system is mainly a software upgrade applied to existing nodes.
The concept was designed according to
3GPP MBMS 23.246 for E-UTRAN and
to coexist with unicast-data and voice
services.
LTE Broadcast gives operators the
flexibility to tailor the way content is
delivered to suit their capabilities.
SFN principles
bandwidth
Maximum usable set of subframes
0 1 2 3 4 5 6 7 8 9 0 1
Subframe = 1ms
Radio frame = 10ms
C1
C2
t
Service dynamics
This supports live streaming and filedelivery use cases. Different service
combinations may be delivered simultaneously over the same bearer.
Cell C1
Cell C2
Sector edge
multipath gain
M2M and B2B
Over the coming decade, machine-tomachine (M2M) data traffic and the
internet of things will create more
connectivity demands on the network
and create the need for diverse types
of eMBMS LTE-enabled devices. LTE
Broadcast technology supports efficient
one-to-many transfer of machine data
in any file format, which can be used for
M2M use cases, off-loading the network
and providing the essential machine
connectivity and control.
Ericsson value proposition
The concept of Ericsson’s LTE Broadcast
solution enables unicast and broadcast
service blending, aiming to help meet
the challenges created by rising mobile
usage and the growth of video traffic in
LTE networks. The solution covers the
entire chain from live encoder, through
delivery via point-to-multipoint transport to devices.
Particular focus has been placed on
the specification and implementation
of the device, starting with the physical
chipset as well as transport control middleware – essential enablers for the creation and deployment of eMBMSs.
Implementing live streaming with
MPEG-DASH4 is a technology choice that
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
supports the common use of a player
on devices and a live encoder head-end
system for both unicast and broadcast
– reducing operating costs and maximizing infrastructure usage. As outlined later in this article, extensive
simulation, lab testing and field trials
have been conducted with the aim of
characterizing the spectral efficiency
of eMBMSs in deployed networks with
mixed traffic profiles.
The results show that live video broadcast with commercially acceptable levels of video and audio degradation is
achievable. For video broadcasting to
smartphones and tablets, compression
using the H.2645 standard is feasible,
with HEVC6 coming sometime in the
near future.
System architecture
Broadcast and unicast radio channels
coexist in the same cell and share the
available capacity. The subset of available radio resources can temporarily be
assigned to a broadcast radio channel.
Mobile-communication systems such
as LTE are traditionally designed for unicast communication, with a separate
radio channel serving each device. The
resources allocated to the device depend
on the data rate required by the service,
Time dynamics
LTE Broadcast activation triggers the
allocation of radio resources on a needs
basis. A session may be active for a short
time say several minutes or for longer
periods: several days in some cases.
When the session is no longer active, the
assigned radio and system resources can
be reallocated for use by other services.
Location dynamics
LTE Broadcast can be activated for small
geographical locations, such as stadiums and city centers, or for large areas,
covering say an entire city or region.
As long as there is sufficient capacity
in the network, multiple broadcast sessions can be active simultaneously.
Resource allocation dynamics
This involves the free allocation of
resources for LTE Broadcast. Up to 60
percent of the FDD radio resources and
up to 50 percent for TDD can be assigned
to a broadcast transmission.
Principles of the radio interface
The LTE radio interface is based on
OFDM in the downlink, where the frequency selective wideband channel
is subdivided into narrowband channels orthogonal to each other. In time
domain, a 10ms radio frame consists
of subframes of 1ms each; where a subframe is the smallest unit with full frequency domain that can be allocated to
a broadcast transmission.
With eMBMS, all users within the
7
broadcast area, provided they have the
right subscription level and an MBMScapable device, can receive broadcasted content. By setting up a single bearer
over the radio interface, operators can
distribute a data stream to an unlimited number of users.
Although it is possible to deliver
broadcasts within a single cell, the concept becomes truly interesting with
SFN, the principles of which are illustrated in the lower part of Figure 2.
Broadcast data is sent over synchronized SFN – tightly synchronized,
identical transmissions from multiple
cells, using the same set of subframes
and modulation and coding schemes,
appear to the device as a transmission
from a single large cell over a time-dispersive channel. This improves received
signal quality and spectral efficiency (as
shown in Figure  2). For a more detailed
description, refer to LTE/LTE-Advanced
for Mobile Broadband7.
The maximal usable set of subframes
is shown in the top left of the diagram,
and the nodes are time-synchronized to
a high precision.
By using long data-symbol duration
in OFDM, it is possible to mitigate the
effect of inter-symbol interference (ISI)
caused by delayed signals. For additional protection against propagation
delays LTE/OFDM uses a guard interval
– delayed signals arriving during the
guard interval do not cause ISI and so
the data rate can be maintained. For
SFN, unlike unicast, signals arrive from
many geographically separate sources
and can incur large delay spread.
Consequently, one of the factors limiting MBMS capacity is self-interference
from signals from transmitters with
a delay that is greater than the guard
interval (low transmitter density). To
overcome this, a long cyclic prefix is
added to MBSFN-reserved subframes
to allow for the time difference in the
receiver and corresponds to an ISD of
approximately 5km.
Architecture
The eMBMS architecture, shown in
Figure  3, is designed to handle transmission requirements efficiently.
The Broadcast Multicast Service
Center (BM-SC) is a new network element at the heart of the LTE Broadcastdistribution tree. Generic files or
FIGURE 3 Architecture – with only eMBMS components shown
Unicast
S1-U
MME
S11
M3 / S1-MME
S/PDN-GW
eNB
SGi
Sm
SGmb
eNB
SGi-mb
M1
eNB
BM-SC
MBMS-GW
Control
User data
MPEG-DASH live video streams are
carried as content across the BM-SC
and made available for broadcast. The
BM-SC adds resilience to the broadcast
by using AL-FEC – which adds redundancy to the stream so that receivers
can recover packet losses – and supports
the 3GPP-associated delivery procedures. These procedures include unicast base file repair – allowing receivers
to fetch the remaining parts of a file
through unicast from the BM-SC and
reception reporting, so operators can
collect QoE reports and make sessionquality measurements.
Another new network element is the
MBMS-GW, which provides the gateway
function between the radio and service
networks. It forwards streams from the
BM-SC to all eNBs participating in the
SFN transmission. IP multicast is used
on the M1 interface between the gateway and the eNBs, so that the packet replication function of existing routers can
be used efficiently. The gateway routes
MBMS session control signaling to the
MMEs serving the area. The MMEs in
turn replicate, filter and forward session control messages to the eNBs participating in the specific broadcast session.
The eNBs provide functionality for
configuration of SFN areas, as well
Content
as broadcasting MBMS user data and
MBMS-related control signaling on
the radio interface to all devices. Note,
the eNB contains the 3GPP Multicell
Coordination Entity (MCE) function.
eMBMS LTE-enabled devices are
an essential part of the ecosystem.
LTE capabilities are becoming integrated into more and more types of
devices and may be implemented on
devices other than phones and tablets
such as embedded platforms for M2M
communications.
The UE platform is divided into three
main blocks (see Figure  4):
the lower block incorporates the LTE
radio layers, which are typically
implemented in the LTE chipset,
supporting unicast as well as broadcast;
the middleware block handles the FLUTE
protocol8 , AL-FEC decoding, unicast file
repair and other functions. It includes
transport control functions, such as
service scheduling, as well as a cache for
post-broadcast file processing; and
the top platform block exposes APIs to
the middleware and connectivity layer
methods.
Application development is enabled
through an SDK, which provides
the platform APIs. The SDK enables
developers to create and test
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Capture your audience
8
FIGURE 4 UE and SDK in eMBMS ecosystem
SDK
App
download
Platform APIs
eMBMS
middleware
LTE chipset
(L1, L2, L3)
Content
SGi-mb
BM-SC
User
equipment
eMBMS-enabled applications without requiring detailed knowledge of the
underlying transport, control, or radiobearer technology.
Spectral efficiency
According to 3GPP specifications,
eMBMSs and unicast services should
be provisioned on a shared frequency.
Consequently, while a broadcast service
is active, radio-interface resources can
be borrowed from unicast capacity.
FIGURE 5 SGmb
LTE
Broadcast
and unicast
network
Spectral efficiency can be defined
as the possible information rate transmitted over a given bandwidth with a
defined loss rate. The information loss
rate depends on the modulation and
coding scheme used for physical transmissions and the protection offered by
AL-FEC. This definition of spectral efficiency includes packet overheads, such
as AL-FEC redundancy.
The simulation results from an evaluation of spectral efficiency are shown in
Evaluating spectral efficiency
Spectral efficiency (b/s/Hz)
3.0
—— indoor; - - - in-car
w/o AL-FEC
AL-rBLER=1e-3
AL-rBLER=1e-5
2.5
2.0
1.5
1.0
0.5
0
0
1
2
3
4
5
6
7
8
9
ISD (km)
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
BOX B  
Standards­
The standard­
ization of MBMS
started in 3GPP
with Rel-6,
which supported
GERAN and
UTRAN access
networks. Over
time, 3GPP has
improved the
access network
support by,
for example,
defining the
integrated
mobile broad­
cast (IMB)
solution, which
uses UTRAN
TDD bands
to offer up to
512kbps per
content channel.
Support for
E-UTRAN access
(LTE) was
added to 3GPP
Rel-9 as part
of the eMBMS
standardization
activity.
Figure  5. The results associated with a
broadcast transmission depend on the
ISD in a link budget – signal-to-noise
ratio (SNR) – limited deployment. Two
urban environments were simulated:
indoor scenarios with 20dB penetration
loss and in-car scenarios with 6dB loss
assuming 95 percent coverage probability in all cases.
The failure criterion used was 10 -3
BLER (corresponding to a packet loss of
four packets per hour) and simulations
were run with and without AL-FEC. An
ideal Raptor code with FEC covering 2s
per source block was used in this evaluation. The payload for each source block
consisted of 50 packets with each IP
packet spanning two transport blocks.
The MBSFN simulation area ­-included
19 sites, each with three sectors. The
results show that MBMS spectral efficiency of about 1-3bps/Hz (indoor/in-car)
could be achieved for a cellular ISD of up
to 2km. The simulation results and additional testing show that FEC improves
video quality and saves capacity.
From the graphs in Figure  5, it is possible to conclude that when ISD is less
than 1km, spectral efficiency is greater than 2.5b/s/Hz. By allocating one subframe for MBMS transmission in 20MHz
spectrum, corresponding to 10 percent
of capacity, the achievable data rate is in
the range of 5Mbps. Live video and file delivery
The two main eMBMS use cases are live
streaming and on-request file delivery.
Live streaming supports services for
real-time video and audio broadcasting, and on-request file delivery enables
services such as unicast off-load (local
device caching), software updates and
M2M file loading. In fact, any arbitrary
file or sequence of files can be distributed over eMBMSs.
The target broadcast area for these
use cases may be any desired size – some
scenarios require a small broadcast
area, such as a venue or a shopping mall,
and other cases require much larger
areas, even up to nationwide coverage.
Ericsson has selected MPEG-DASH for
live streaming delivery over eMBMSs.
This solution slices the live stream into
a sequence of media segments, which
are then delivered through the system
as independent files.
Typically, HTTP is used to fetch these
9
files. In the eMBMS case, one quality representation is delivered as a sequence of
files through eMBMSs using MBMS file
delivery.
By using MPEG-DASH with eMBMSs,
the same live encoder and common
clients can be used for unicast and
broadcast offerings. This solution also
supports using the same system protocol stack for both live streaming and
file-delivery implementation.
The IETF FLUTE protocol8 allows distribution of files over unidirectional
links using UDP. Most service-layer features can be used for both streaming
and file delivery; transmission reliability can be increased using AL-FEC in both
cases. File delivery can also make use of
the unicast file-repair feature – allowing
UEs to fetch any missing file segments.
However, this feature is not intended
for use with services that have real-time
requirements, such as live streaming.
With FLUTE, delivery and eMBMS
sessions are used, where the duration
of a delivery session may span one or
more eMBMS sessions. The broadcast
is active for the entire eMBMS session,
during which UEs can receive content.
The relationship between delivery sessions and eMBMS sessions is shown in
Figure  6. Service announcement is
used to inform devices about delivery
sessions and also about eMBMS sessions
using a schedule description. UEs do not
need to monitor the radio interface for
eMBMS sessions continuously.
In Figure  6, the schedule description
instructs the UE to expect an eMBMS
session between t2 and t3 and between t6
and t7. Before the UE expects an eMBMS
session, it is already active on the radio
interface (t1 < t2). When it comes to filedelivery services, it is preferred that
devices should search for sessions prior
to expected transmission time on the
radio, to ensure that they do not miss
the start of a transmission.
The example in Figure  6 could represent a service, such as downloading an
application that allows users to activate,
receive and interact with the broadcast
using unicast services from a phone, tablet or television.
From the point of view of the user
and the UE middleware, the two broadcasts belong to the same MBMS user service, which presents a complete offering
including activation and deactivation.
FIGURE 6 Example of two scheduled broadcasts
Delivery session (FLUTE)
Service announcement
Service announcement
informs the UE about
the schedule
Service announcement
informs the UE about
the schedule
eMBMS session
eMBMS session
t1 t2
t 3 t4
t5 t6
t6 to t7
UEs expects to receive data
of that FLUTE session
UEs expects to receive data
of that FLUTE session
is used to distribute the content over
the air interface.
LTE Broadcast provides operators
with techniques to deliver consistent
service quality, even in highly crowded
areas. Such techniques for delivering
content efficiently are valuable as they
free up capacity, which can be used for
other services and voice traffic.
eMBMS architecture
Application
server
App
FR/RR (HTTP)
SGW
S1-U
UE
Time
t2 to t3
Conclusions
The data volume in mobile networks is
booming mostly due to the success of
smartphones and tablets.
LTE Broadcast is one way of providing new and existing services in areas
that can at times be device dense, such
as stadiums and crowded city centers.
Single-frequency network technology
FIGURE 7 t7 t8
Uu
S1-MME
eNB
S5/S8
BM-SC
PGW
M3
CDN/
live encoder
SGi
S11
MME
HTTP
SGmb
Sm
MBMS-GW
SGi-mb
M1
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Capture your audience
10
References
1. Ericsson, 2012, Ericsson ConsumerLab report, TV
and video – changing the game, available at: http://
www.ericsson.com/res/docs/2012/consumerlab/
consumerlab-tv-video-changing-the-game.pdf
2.Ericsson, November 2012, Ericsson Mobility Report, On the
pulse of the Networked Society, available at: http://www.
ericsson.com/res/docs/2012/ericsson-mobility-reportnovember-2012.pdf
3.Mobile Content Venture, June 2012, Dyle Mobile TV Data
Report, available at: http://www.dyle.tv/assets/Uploads/
DyleReport.pdf
4.ISO/IEC 23009-1:2012, Information technology – Dynamic
adaptive streaming over HTTP (DASH) — Part 1: Media
presentation description and segment formats, available
at: http://www.iso.org/iso/iso_catalogue/catalogue_tc/
catalogue_detail.htm?csnumber=57623
5.ITU-T H.264 Advanced video coding for generic audiovisual
services, available at: http://www.itu.int/rec/T-REC-H.264
6.ITU-T H.265 / ISO/IEC 23008-2 HEVC, available
at: http://www.itu.int/ITU-T/aap/AAPRecDetails.
aspx?AAPSeqNo=2741
7.Erik Dahlman, Stefan Parkvall, Johan Sköld,
2011, LTE/LTE-Advanced for Mobile Broadband,
available at: http://www.elsevier.com/books/4glte-lte-advanced-for-mobile-broadband/
dahlman/978-0-12-385489-6
8.IETF RFC 3926, FLUTE – File delivery over unidirectional
transport, T. Paila, et al., October 2004, available at: http://
tools.ietf.org/html/rfc3926
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Thorsten Lohmar
Michael Slssingar
joined Ericsson in
Germany in 1998 and
worked in different
Ericsson Research units
for several years. He worked on a
variety of topics related to mobilecommunication systems and led
research projects specifically in the
area of multimedia technologies. On
the development front, he is focusing
on the technical coordination of
eMBMSs with an end-to-end
perspective. He is currently working as
a senior specialist for end-to-end video
delivery, principally in mobile networks.
Lohmar holds a Ph.D. in electrical
engineering from RWTH Aachen
University, Germany.
is an Ericsson senior
specialist in service
delivery architectures and
holds a post-graduate
diploma and master’s in computing and
software engineering. He has held
many senior engineering roles at
Ericsson, mainly in the media-delivery
field, and has contributed to the
Ericsson IPTV and Mobile TV delivery
solutions. In the field of MBMS,
Slssingar initially specialized in WCDMA
MBMS, where he helped develop the
Ericsson Content Delivery System.
More recently, he has worked with LTE
eMBMS broadcast, where he has a
strong interest in the service layer BMSC node, UE middleware and metadata
provisioning areas.
Stig Puustinen
Vera Kenehan
is a senior project
manager at System
Management within
Business Unit Networks,
where he is currently running an LTE/
EPC systems project involving
extensive eMBMSs work. He joined
Ericsson in 1991, and has since held a
variety of project- and programmanagement roles. He was involved in
the early releases of GSM, the first
introduction of WCDMA/HSPA and the
first release of LTE/EPC.
is a strategic product
manager within LTE Radio
and has worked with
several generations of
radio-access technologies, including
LTE, WCDMA and PDC. She was largely
involved in the initial standardization of
LTE, including eMBMS. For the past two
years, she has been working on the
MBMS product as well as promoting
and bringing eMBMS to the market.
She holds a master’s in telecom
engineering from the University of
Belgrade, Serbia.
Re:view
11
Nine decades
of innovation
The Roaring Twenties
In 1924, in the very first
issue of Ericsson Review, the
editor stated that the
objective of the magazine
was to take up points of
design and construction,
which had not yet reached
final standardization, for
discussion. The cover article
featured Ericsson’s presence
at the Gothenburg Exhibition
(1923), where a giant replica
of our standard table set
telephone housed a complete Ericsson Review,
issues 1 & 2, 1924.
Ericsson exchange for 500
lines, to which a few
telephones were connected. Visitors could make
calls and watch the switching process, through the
plate-glass windows.
Automatic exchanges to smart networks
The history of our technology is deeply entrenched in that
of the telecoms industry. War, international terrorism, and
developments in other industries such as railways and the
more recent digital revolution have shaped our playing field.
Our technical expertise is not just based on theory; it is
about how to apply the right technology to create commercially viable products. Standards – and the ability to create
and evolve them – have helped us and our industry become
global providers of interoperable solutions – a cornerstone of
the Networked Society.
Over the past nine decades, the world has probably changed
more than ever before. Here are a few highlights from
Ericsson’s history and some of the world events that have
played their part in the evolution of telecoms.
In 1924, Ericsson launched its automatic 500-point system. Highlighted in the very first issue of Ericsson Review,
this revolutionary switching system was demonstrated at
the Gothenburg Exhibition of 1923. In hindsight, it sounds
curious that attendees could ‘watch’ the switching process,
which at the time was mechanical. By the 1930s, Ericsson had
delivered about 100 systems with a total of more than 350,000
lines. Sales of the system continued to rise over the coming
decades, not declining significantly until the 1970s. By 1974,
4.8 million lines using this system were in operation in public telephone stations.
The very first 500-point system was put into service at
the Norra Vasa exchange in Stockholm, and was still in
operation 60 years later. At the time of installation in 1924,
Televerket – the Swedish government agency for telecommunications (1853-1993) – had four different systems to choose
from, including a crossbar system also developed by Ericsson.
Televerket’s choice was a decisive one for the future development of Ericsson.
In the 1920s and 1930s, the opening of a new telephone
exchange was a major event for small towns and villages.
Official opening ceremonies were often carried out by the
local mayor, to the backdrop of a brass band and refreshments
provided for the locals by the operator and the vendor.
In the 1930s, Ericsson introduced a photoelectric announcing machine – a simple device that could deliver short prerecorded messages. This offloaded the work of operators and
had the ability to deliver longer announcements, such as the
speaking clock and – later in the decade – automatic weather
forecasts. The first device for weather forecasts was put into
service in Stockholm on June 1, 1936 and was the first of its
kind in the world. It wasn’t until two decades later that they
were replaced with newer technologies.
In the 1940s, Ericsson introduced an information management structure called the ABC System. Ericsson still uses this
system to classify and manage its information and products.
Depression and modernism
The cover of issue 2, 1934
shows the ‘photo-electric
talking machine for
automatic time indication.’
This issue included an
account of a carrier system
supplied by Ericsson to the
Indian Radio and
Communications Company.
In the spirit of modernism,
this issue included a number
of articles relating to the use
of clocks and timing
Ericsson Review,
mechanisms in industry. An
issue 2, 1934.
article on party lines
discussed the problem of
connecting several telephone instruments to one line. This
was of particular interest at the time, as exchanges in rural
districts were being automated and party lines were in need
of some degree of modification. One of the main concerns
was how to use party lines without altering the subscriber
equipment.
The war years
Ericsson Review was not published in English between
1940 and 1944 due to the ongoing world war. Apart from
the fact that the paper situation necessitated a reduction
both of quantity and quality of printing paper, the journal
was issued as usual in its Swedish edition during the war
years. In 1945, a collection of some of the articles published
during the war in Swedish were printed in a composite
English language edition.
Continued on
page 19...
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Dispelling the NLOS myths
12
Non-line-of-sight microwave
backhaul for small cells
The evolution to denser radio-access networks with small cells in cluttered
urban environments has introduced new challenges for microwave backhaul.
A direct line of sight does not always exist between nodes, and this creates a need
for near- and non-line-of-sight microwave backhaul.
JONA S H A N S RY D, JONA S E D S TA M , BE NGT-E R I K OL S S ON A N D C H R I S T I NA L A R S S ON
Using non-line-of-sight (NLOS)
propagation is a proven approach
when it comes to building
RANs However, deploying
high-performance microwave
backhaul in places where there
is no direct line of sight brings
new challenges for network
architects. The traditional belief
in the telecom industry is that
sub-6GHz bands are required
to ensure performance for
such environments. This article
puts that belief to the test,
providing general principles,
key system parameters and
simple engineering guidelines for
deploying microwave backhaul
using frequency bands above
20GHz. Trials demonstrate that
such high-frequency systems
can outperform those using sub6GHz bands – even in locations
with no direct line of sight.
Point-to-point microwave is a cost-­
efficient technology for flexible and
rapid backhaul deployment in most
locations. It is the dominant backhaul
medium for mobile networks, and is
expected to maintain this position as
mobile broadband evolves; with microwave technology that is capable of providing backhaul capacity of the order of
several gigabits-per-second1.
BOX A Complementing the macro-cell layer by
adding small cells to the RAN introduces new challenges for backhaul. Smallcell outdoor sites tend to be mounted
3-6m above ground level on street fixtures and building facades, with an
inter-site distance of 50-300m. As a large
number of small cells are necessary to
support a superior and uniform user
experience across the RAN2, small-cell
backhaul solutions need to be more costeffective, scalable, and easy to install
than traditional macro backhaul technologies. Well-known backhaul technologies such as spectral-efficient LOS
microwave, fiber and copper are being
tailored to meet this need. However,
owing to their position below roof
height, a substantial number of small
cells in urban settings do not have access
to a wired backhaul, or clear line of sight
to either a macro cell or a remote fiber
backhaul point of presence.
The challenges posed by locations
without a clear line of sight are not new
to microwave-backhaul engineers,
who use several established methods
to overcome them. In mountainous
terrain, for example, passive reflectors
and repeaters are sometimes deployed.
However, this approach is less desirable for cost-sensitive small-cell backhaul, as it increases the number of sites.
In urban areas, daisy chaining is often
used to reach sites in tricky locations – a
Terms and abbreviations
FDD
frequency division duplexing
LOSline-of-sight
MIMO multiple-input,
multiple-output
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
NLOSnon-line-of-sight
OFDM
orthogonal frequency
division multiplexing
RAN
radio-access network
TDD
time division duplexing
solution that is also effective for smallcell backhaul (see Figure 1).
Network architects aim to dimension backhaul networks to support
peak cell-capacity3 – which today can
reach 100Mbps and above. However, in
reality, there is a trade-off among cost,
capacity and coverage resulting in a
backhaul solution that, at a minimum,
can support expected busy-hour traffic
with enough margin to account for statistical variation and future growth: in
practice around 50Mbps with availability requirements typically relaxed to
99-99.9 percent. Such availability levels
require fade margins of the order of just
a few decibels for short-link distances.
For small-cell backhaul simplicity
and licensing cost are important issues.
Light licensing or technology-neutral
block licensing are attractive alternatives to other approaches such as link
licensing, as they provide flexibility4.
Using unlicensed frequency bands can
be a tempting option, but may result in
unpredictable interference and degrad­
ed network performance. The risk
associated with unlicensed use of the
57-64GHz band is lower than that associated with the 5.8GHz band, owing to
higher atmospheric attenuation, sparse
initial deployment, and the possibility
of using compact antennas with narrow beams, which effectively reduces
interference.
Providing coverage in locations without a clear line of sight is a familiar part
of the daily life of mobile-broadband
and Wi-Fi networks. However, maybe
because such locations are commonplace, a number of widespread myths
and misunderstandings surrounding
NLOS microwave backhaul exist – for
example, that NLOS microwave backhaul needs sub-6GHz frequencies, widebeam antennas and OFDM-based radio
13
technologies to meet coverage and
capacity requirements. Despite this, a
number of studies on NLOS transmission using frequency bands above 6GHz,
for example, have been carried out for
fixed wireless access5 and for mobile
access6. Coldrey et al. showed that it
is realistic to reach 90 percent of the
sites in a small-cell backhaul deployment with a throughput greater than
100Mbps using a paired 50MHz channel at 24GHz7.
FIGURE 1 Microwave backhaul scenarios for small-cell deployment
Daisy chain
Penetration
NLOS principles
As illustrated in Figure 1, all NLOS propagation scenarios make use of one or
more of the following effects:
Diffraction
Fiber
diffraction;
reflection; and
penetration.
All waves change when they encounter
an obstacle. When an ­electromagnetic
wave hits the edge of a building, diffraction occurs – a phenomenon often
described as the bending of the signal.
In reality, the energy of the wave is scattered in the plane perpendicular to the
edge of the building. The energy loss –
which can be considerable – is proportional to both the sharpness of the bend
and the frequency of the wave8.
Reflection, and in particular random
multipath reflection, is a phenomenon
that is essential for mobile broadband
using wide-beam antennas. Single-path
reflection using narrow-beam antennas
is, however, more difficult to engineer
owing to the need to find an object that
can provide the necessary angle of incidence to propagate as desired.
Penetration occurs when radio waves
pass through an object that completely
or partially blocks the line of sight. It is
a common belief that path loss resulting from penetration is highly dependent on frequency, which in turn rules
out the use of this effect at higher frequencies. However, studies have shown
that in reality path loss due to penetration is only slightly dependent on frequency, and that in fact it is the type and
thickness of the object itself that creates
the impact on throughput9, 10. For example, thin, non-metallic objects – such as
sparse foliage (as shown in Figure 1) –
add a relatively small path loss, even for
high frequencies.
Deployment guidelines can be
defined given a correct understanding
Reflection
and application of these three propagation effects, giving network engineers
simple rules to estimate performance
for any scenario.
System properties
A simplified NLOS link budget can be
obtained by adding an NLOS path attenuation term (ΔLNLOS) to the traditional
LOS link budget, as shown in Equation 1.
Equation 1
PRX = PTX + GTX + GRX - 92 - 20log(d) - 20log(f) - LF - ΔLNLOS
Here, PRX and PTX are the received and
transmitted power (dBm – ratio of power in decibels to 1 milliwatt); GTX and
GRX are antenna gain (in decibels isotropic – dBi) for the transmitter and receiver respectively; d is the link distance
(in kilometers); f is the frequency (in
gigahertz); LF is any fading loss (in decibels); and ΔLNLOS is the additional loss
(in decibels) resulting from the deployment of NLOS-propagation effects. Not
shown in this equation is the theoretical frequency dependency of the antenna gain, which for a fixed antenna size
will increase as 20log(f) and as a consequence, the received signal – PRX – will
actually increase as 20log(f) when carrier frequency is increased for a fixed
antenna size. This relationship indicates the advantage of using higher frequencies for applications where a small
antenna form factor is of importance
– as is the case for small-cell backhaul.
To determine the importance of
NLOS-system properties, Ericsson carried out measurement tests on two commercially available microwave backhaul
systems in different frequency bands
(described in Table 1). The first system
used the unlicensed 5.8GHz band with
a typical link configuration for applications in this band. The air interface used
up to 64QAM modulation in a 40MHzwide TDD channel with a 2x2 MIMO
(cross-polarized) configuration providing full duplex peak throughput of
100Mbps (200Mbps aggregate). The second system, a MINI-LINK PT2010, used
a typical configuration for the licensed
28GHz band, based on FDD, 56MHz
channel spacing and single-carrier
technology with up to 512QAM modulation, providing full duplex throughput of 400Mbps (800Mbps aggregate).
To adjust the throughput based on the
quality of the received signal, both
Table 1: Test system specifications
SYSTEM
TECHNOLOGY
CHANNEL
SPACING
ANTENNA
GAIN
OUTPUT
POWER
PEAK
THROUGHPUT
5.8GHz
TDD/OFDM
64QAM
40MHz
17dBi
19dBm
100Mbps
28GHz
FDD/single carrier
512QAM
56MHz
38dBi
19dBm
400Mbps
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Dispelling the NLOS myths
14
FIGURE 2 advantages of using higher frequencies
are clear: with comparable antenna sizes, the link margin is about 20dB higher
at a peak rate of 400Mbps for the 28GHz
system compared with the 5.8GHz system at a peak rate of 100Mbps.
Link margin as a function of throughput and distance
Link margin (dB)
80
28GHz
5.8GHz
70
60
90Mbps
185Mbps
50
280Mbps
10Mbps
40
400Mbps
60Mbps
30
80Mbps
100Mbps
20
10
50
100
150
200
250
300
350
400
450
500
Link distance (m)
systems used adaptive modulation.
Physical antenna sizes were similar, but
due to the frequency dependency of the
antenna gain and the parabolic type
used in the 28GHz system, it offered a
gain of 38dBi while the flat antenna of
the 5.8GHz system reached 17dBi.
Link margin versus throughput and
hop distance is shown in Figure 2.
FIGURE 3A Here, the margin is defined as the
difference between received power
(according to Equation 1) and the receiver threshold for a particular modulation level (throughput) – in line of sight
conditions without fading (Lf = 0). If
ΔLNLOS caused by any NLOS effect can
be predicted, the curves in Figure 2 can
be used to estimate throughput. The
Test site for NLOS backhaul – diffraction
0m
15
© 2013 BLOM © 2013 Microsoft Corporation
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Measurements
Diffraction
It is commonly believed that the diffraction losses occurring at frequencies
above 6GHz are prohibitively high, and
consequently, deploying a system using
this effect for NLOS propagation at such
frequencies is not feasible. However,
even if the absolute loss can be relatively high, 40dB and 34dB for the 28GHz
and 5.8GHz systems respectively (with
a diffraction angle of 30 degrees), the
relative difference is only 6dB8 – much
less than the difference in gain for comparable antenna sizes even when taking into account the higher free-space
loss for the 28GHz system (see Figure 2).
Figures 3A and 3B show the setup and measured results of a scenario
designed to test diffraction. A first radio
was positioned on the roof of an office
building (marked in Figure 3A with a
white circle). A second radio was mounted on a mobile lift, placed 11m behind
a 13m-high parking garage. The effect
on the signal power received by the second radio was measured by lowering the
mobile lift. Figure 3B shows the measured received signal-power versus distance below the line of sight for both
test systems, as well as the theoretical
received power calculated using the ideal knife-edge model8. Both radios transmitted 19dBm output power, but due
to the 21dBi lower antenna gain for the
5.8GHz system, the received signal for
this radio was 20dB weaker after NLOS
propagation than the 28GHz system.
The measured results compare well
against the results based on the theoretical model, although an offset of a couple
of decibels is experienced by the 28GHz
system – a small deviation that is expected due to the simplicity of the model.
To summarize, diffraction losses
can be estimated using the knife-edge
­model8. However, due to the model’s
simplicity, losses calculated by it are
slightly underestimated. This can be
compensated for in the planning process by simply adding a few extra decibels to the loss margin.
15
The 28GHz system can sustain full
throughput at much deeper NLOS
than the 5.8GHz system, which is to be
expected as it has a higher link margin. Full throughput – 400Mbps – was
achieved at 28GHz up to 6m below the
line of sight, equivalent to a 30-degree
diffraction angle, while the 5.8GHz system dropped to under 50Mbps at 3m
below the line of sight. The link margin is the single most important system
parameter for NLOS propagation and, as
expected, the 28GHz system performs
in reality better in a diffraction scenario than a 5.8GHz system with comparable antenna size.
Reflection
The performance characteristics of the
5.8GHz and 28GHz systems were measured in a single-reflection scenario in
an area dominated by metal and brick
facades – shown in Figure 4A. The
first radio was located on the roof of the
office building (marked with a white
circle), 18m above ground level; and the
second on the wall of the same building, 5m above ground, facing the street
canyon. The brick facade of the building
on the other side of the street from the
second radio was used as the reflecting
object, resulting in a total path length of
about 100m. The reflection loss will vary
with the angle of incidence, which in
this case was approximately 15 degrees,
resulting in a ΔLNLOS of 24dB for the
28GHz system and 16dB for the 5.8GHz
system – figures that are in line with earlier studies11. Reflection loss is strongly
dependent on the material of the reflecting object, and for comparison purposes ΔLNLOS for a neighboring metal facade
was measured to be about 5dB for both
systems with similar angle of incidence.
To summarize, it is possible to c­ over
areas that are difficult to reach using
multiple reflections in principle.
However, taking advantage of more
than two reflections is in practice problematic – due to limited link margins
and the difficulty of finding suitably
aligned reflection surfaces. ΔLNLOS predictions for a single-facade reflection
in the measured area can be expected to vary between 5dB and 25dB at
28GHz and between 5dB and 20dB at
5.8GHz. The throughput for both systems measured over 16 hours is shown
in Figure 4B.
FIGURE 3B Throughput and received power – diffraction
Throughput (Mbps)
Received power (dBm)
700
-10
Received power 28GHz
Received power 5.8GHz
Theoretical power 28GHz
Theoretical power 5.8GHz
Throughput 28GHz
Throughput 5.8GHz
600
500
-20
-30
400
-40
300
-50
200
-60
100
-70
0
–2
-80
0
2
4
6
8
10
Distance below line of sight (m)
The 28GHz system shows a stable
throughput of 400Mbps, while the
throughput for the 5.8GHz system, with
a much wider antenna beam, dropped
from 100Mbps to below 70Mbps. These
variations are to be expected owing
to the fact that the wider beam experiences a stronger multipath. OFDM
is an effective mitigation technology that combats fading, which will,
FIGURE 4A at severe multipath fading, result in
a graceful degradation of throughput
– as illustrated. However, the narrow
­antenna lobe at 28GHz, in combination with the advanced equalizer of the
high-­performance MINI-LINK radio,
effectively suppresses any multipath
degradation, enabling the use of a single-carrier QAM technology for NLOS
conditions – even up to 512QAM and
56MHz channel bandwidths.
Test site for NLOS backhaul – reflection
100m
© 2012 TerraItaly © 2013 Microsoft Corporation
Pictometry Bird’sEye © 2012 Pictometry International Corp.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Dispelling the NLOS myths
16
FIGURE 4B up to more than 28dB. A complementary experiment showing similar excess
path loss was carried out at 5.8GHz.
To summarize, contrary to popular
belief, a 28GHz system can be used with
excellent performance results using
the effect of NLOS penetration through
sparse greenery.
Throughput over time – single reflection
Throughput (Mbps)
28GHz
5.8GHz
420
400
380
120
100
80
60
0
2
4
6
8
10
12
14
16
Time (hours)
Penetration
As with the case for NLOS reflection,
the path loss resulting from penetration is highly dependent on the material of the object blocking the line of
sight. The performance of both test
systems was measured in a scenario
shown in Figures 5A and 5B. The sending and receiving radios were located
150m apart, with one tall sparse tree
and a shorter, denser tree blocking the
line of sight. The radio placed on the
mobile lift was positioned to measure
the radio beam first after penetration
of the sparse foliage and then lowered
to measure the more dense foliage, as
FIGURE 5A shown in Figure 5A. The circle and triangle symbols indicate where the radio
beams exit the foliage.
Measurements were carried out
under rainy and windy weather conditions, resulting in variations of the
NLOS path attenuation, as shown in the
received signal spectra for the 28GHz
radio link in Figure 5B. Under LOS conditions the amplitude spectrum envelope reached -50dBm. Consequently,
the excess path loss for the single-tree
(sparse foliage) scenario varied between
0 and 6dB when measured for 5 minutes. In the double-tree (dense foliage)
case excess path loss varied from 8dB
Test site for NLOS backhaul – penetration
= sparse foliage
= dense foliage
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Deployment guidelines
So far, this article has covered the key
system properties for NLOS propagation – diffraction, reflection and penetration – dispelling the myth that these
effects can be used only with sub-6GHz
frequencies. The next step is to apply the
theory and the test results to an ­actual
deployment scenario for microwave
backhaul.
Table 2 shows the indicative throughput for each NLOS scenario, using the
measured loss from the examples above
together with the graphs in Figure 2.
A trial site, shown in Figure 6, was
selected to measure the coverage for an
NLOS backhaul deployment ­scenario.
Four- to six-story office buildings with
a mixture of brick, glass and metal
facades dominate the trial area. The hub
node was placed 13m above ground on
the corner of a parking garage at the
south end of the trial area. By using the
measured loss in the diffraction, reflection and penetration from the tests as a
rule of thumb, an indicative throughput for each NLOS scenario has been
taken from Figure 2 and summarized
in Table 2.
The colored areas in Figure 6 show the
line of sight conditions for the trial site:
the green areas show where pure LOS
exists; the yellow areas indicate the use
of single-path reflection; the blue areas
indicate diffraction; and the red areas
show where double reflection is needed. Areas without color indicate either
that no throughput is expected or that
they are outside the region defined for
measurement. Measurements were
made within the region delineated by
the dashed white lines. Referring to
Table 2, it is expected that the 5.8GHz
system will meet small-cell backhaul
requirements (>50Mbps throughput)
within a 250m radius of the hub; and
the 28GHz system should provide more
than 100Mbps full duplex throughput
up to 500m from the hub. To test the
actual performance, a receiver node
17
was placed 3m above ground measuring
full duplex throughput along the main
street canyon and in the neighboring
streets. On account of the wide antenna
lobe of the 5.8GHz system, realignment
was not needed for the hub antenna for
measurement purposes. For the 28GHz
system, realignment of the narrow
antenna beam was needed at each measurement point – a fairly simple procedure even under NLOS conditions.
The actual values observed at each
measurement point exceeded or
matched the predicted performance levels in Table 2. Due to the lack of correctly aligned reflection surfaces, providing
backhaul coverage using the doublereflection technique (the red areas of
the trial area in Figure 6) was only possible for a limited set of measurements.
Multipath propagation, including
the reflection effects created by vehicles moving along the street canyon,
was significant for the 5.8GHz system,
but resulted only in slightly reduced
throughput in some of the more difficult scenarios for the 28GHz system.
Summary
In traditional LOS solutions, high system gain is used to support targeted link
distance and mitigate fading caused by
rain. For short-distance solutions, this
gain may be used to compensate for
NLOS propagation losses instead. Sub6GHz frequency bands are proven for
traditional NLOS usage, and as shown
in this article, using these bands is a
viable solution for small-cell backhaul.
However, contrary to common belief,
but in line with theory, MINI-LINK
microwave backhaul in bands above
20GHz will outperform sub-6GHz systems under most NLOS conditions.
The key system parameter enabling
the use of high-frequency bands is the
much higher antenna gain for the same
antenna size. With just a few simple
engineering guidelines, it is possible to
plan NLOS backhaul deployments that
provide high network performance.
And so, in the vast amount of dedicated spectrum available above 20GHz,
microwave backhaul is not only capable of providing fiber-like multi-gigabit
capacity, but also supports high performance backhaul for small cells, even in
locations where there is no direct line
of sight.
FIGURE 5B Channel amplitude response – penetration
Sparse foliage
Dense foliage
Received power (dBm)
Received power (dBm)
–40
–40
Maximum power over 5 minutes
Minimum power over 5 minutes
–50
Maximum power over 5 minutes
Minimum power over 5 minutes
–50
–60
–60
~6dB fluctuation
>20dB fluctuation
–70
–80
29.32
–70
29.34
29.36
29.38
29.40
29.42
–80
29.32
29.34
29.36
29.38
Frequency (GHz)
29.40
29.42
Frequency (GHz)
Resolution bandwidth: 50kHz
Table 2: Indicative bitrate performance for different NLOS key scenarios
0-100m
5.8GHz
28GHz
LOS
SINGLE
REFLECTION
DOUBLE
REFLECTION
DIFFRACTION*
PENETRATION***
100Mbps
100Mbps
10Mbps**
80Mbps
100Mbps
100-250m
100Mbps
80Mbps
<10Mbps**
60Mbps
100Mbps
250-500m
100Mbps
60Mbps
<10Mbps**
10Mbps**
80Mbps
0-100m
400Mbps
400Mbps
280Mbps**
400Mbps
400Mbps
100-250m
400Mbps
400Mbps
185Mbps**
400Mbps
400Mbps
250-500m
400Mbps
400Mbps
185Mbps**
280Mbps
400Mbps
*30-degree diffraction angle; **not recommended for small-cell backhaul; ***sparse foliage or similar
FIGURE 6 NLOS backhaul trial area
Legend
line-of-sight
500m
single reflection
diffraction
double reflection
250m
100m
© 2013 BLOM © 2013 Microsoft Corporation
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Dispelling the NLOS myths
18
References
1. Ericsson, 2011, Ericsson Review, Microwave capacity
evolution, available at: http://www.ericsson.com/res/docs/
review/Microwave-Capacity-Evolution.pdf
2. Ericsson, 2012, White Paper, It all comes back to backhaul,
available at: http://www.ericsson.com/res/docs/
whitepapers/WP-Heterogeneous-Networks-Backhaul.pdf
3. NGMN Alliance, June 2012, White Paper, Small Cell
Backhaul Requirements, available at: http://www.ngmn.
org/uploads/media/NGMN_Whitepaper_Small_Cell_
Backhaul_Requirements.pdf
4. Electronic Communications Committee (ECC), 2012, Report
173, Fixed service in Europe – current use and future trends
post, available at: http://www.erodocdb.dk/Docs/doc98/
official/pdf/ECCRep173.PDF
5. Seidel, S.Y.; Arnold, H.W.; 1995, Propagation measurements
at 28 GHz to investigate the performance of local multipoint
distribution service (LMDS), available at: http://ieeexplore.
ieee.org/xpl/articleDetails.jsp?arnumber=502029
6. Rappaport, T.S.; Yijun Qiao; Tamir, J.I.; Murdock, J.N.;
Ben-Dor, E.; 2012, Cellular broadband millimeter wave
propagation and angle of arrival for adaptive beam steering
systems (invited paper), available at: http://ieeexplore.ieee.
org/xpl/articleDetails.jsp?arnumber=6175397
7. Coldrey, M.; Koorapaty, H.; Berg, J.-E.; Ghebretensaé, Z.;
Hansryd, J.; Derneryd, A.; Falahati, S.; 2012, Small-cell
wireless backhauling: a non-line-of-sight approach for pointto-point microwave links, available at: http://ieeexplore.
ieee.org/xpl/articleDetails.jsp?arnumber=6399286
8. ITU, 2012, Recommendation ITU-R P.526, Propagation
by diffraction, available at: http://www.itu.int/
rec/R-REC-P.526-12-201202-I
9. Anderson, C.R.; Rappaport, T.S.; 2004, In-building
wideband partition loss measurements at 2.5 and 60 GHz,
available at: http://ieeexplore.ieee.org/xpl/articleDetails.
jsp?arnumber=1296643
10.Okamoto, H.; Kitao, K.; Ichitsubo, S.; 2009, Outdoor-toIndoor Propagation Loss Prediction in 800-MHz to 8-GHz
Band for an Urban Area, available at: http://ieeexplore.ieee.
org/xpl/articleDetails.jsp?arnumber=4555266
11.Dillard, C.L.; Gallagher, T.M.; Bostian, C.W.; Sweeney, D.G.;
2003, 28GHz scattering by brick and limestone walls,
available at: http://ieeexplore.ieee.org/xpl/articleDetails.
jsp?arnumber=1220086
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Jonas Edstam
Jonas Hansryd
joined Ericsson in 1995
and is currently head of
technology strategies at
Product Line Microwave
and Mobile Backhaul. He is an expert in
microwave radio-transmission
networks focusing on the strategic
evolution of packet-based mobile
backhaul and RAN. He holds a Ph.D. in
applied solid-state physics from
Chalmers University of Technology,
Gothenburg, Sweden.
joined Ericsson
Research in 2008 and is
currently managing the
microwave high-speed
and electronics group. He holds a Ph.D.
in electrical engineering from the
Chalmers University of Technology,
Gothenburg, Sweden and was a visiting
researcher at Cornell University,
Ithaca, US, from 2003-2004.
Bengt-Erik Olsson
Christina Larsson
joined Ericsson
Research in 2007 to work
on ultra-high-speed
optical communication
systems. Recently he switched interest
to wireless-technology research, and is
currently working on NLOS backhaul
applications for microwave links. He
holds a Ph.D. in optoelectronics from
Chalmers University of Technology,
Gothenburg, Sweden.
joined Ericsson
Research in 2010.
Her current focus area
is microwave backhaul
solutions. She holds a Ph.D. in
electrical engineering from Chalmers
University of Technology, Gothenburg,
Sweden, and was a post-doctoral
researcher at the University of
St. Andrews, St. Andrews, UK, from
2004-2006.
Acknowledgements
The authors gratefully acknowledge the colleagues who have contributed to
this article: Jan-Erik Berg, Mikael Coldrey, Anders Derneryd, Ulrika Engström,
Sorour Falahati, Fredrik Harrysson, Mikael Höök, Björn Johannisson,
Lars Manholm and Git Sellin
Re:view
19
Nine decades
of innovation
Recovery and prosperity In 1954, the cover article
of Ericsson Review issue 1
looked at the traffic reliability
of the crossbar system,
which Ericsson delivered to
the Helsinki Telephone
Corporation in 1950. The
cover illustrates the interior
of the Helsinki exchange for
PBX subscribers. The final
testing of the system was
carried out in the latter half
of 1953 and gave a fault rate
Ericsson Review,
of 0.090 percent based on
issue 1, 1954.
20,000 test connections.
This result was deemed to be
highly satisfactory, as the exchange was very heavily
loaded during peak periods.
Automatic exchanges to smart networks
Toward the end of this decade, Ericsson trialed a crossbar
system, 30 years after these switches were first put into practical operation.
The Second World War had a profound effect on people and
business. Widespread misery, rations and a shortage of raw
materials naturally led many companies to diversify their
operations. But overall, the slowdown in growth and reduced
level of information exchange among researchers led to a dip
in development. World average telephone density in 1930 was
2 per 100 capita; by 1950 this figure had risen to 3, and in the
two decades following that, subscriber density more than doubled – reaching a total of 7 in 1970.
In preparation for the 1952 Summer Olympics, Ericsson
installed its first commercial crossbar system in Helsinki,
Finland, in 1950. The decision to move to crossbar switching
came about during the war as Ericsson was developing smaller exchanges for rural communities and enterprises. This
technology, however, presented new challenges in terms of
traffic engineering and dimensioning in particular. In his thesis on a study of congestion in link systems, lifelong Ericsson
employee Christian Jacobæus presented a way to calculate
traffic capacity that was subsequently trialed and became a
worldwide standard.
Computing was the next wave of technology to be adopted by the telecoms industry. Parts of Ericsson’s AKE range of
telephone exchanges were computer controlled, the key feature being a Stored Program Controlled (SPC) element, which
managed the switches and operated in real time. The initial
commercial deployments in 1968 were the first computercontrolled exchanges outside the US.
By the end of the 1960s, however, it had become clear
that something different was needed. Something that was
more flexible – a modular system that could be expanded to
accommodate new technologies and services without the
need for fundamental system changes: something that was
future-proof.
The downturn of the global economy following the oil crisis in the 1970s again led to several tough years for industry in
general. For Ericsson, a boost came when the Saudi Ministry of
Telecommunications chose Ericsson AXE exchanges in what
was to be the biggest contract in the history of telecoms at that
time. This digital exchange introduced modular design and
became one of the world’s most successful switching systems.
In 1984, Ericsson Review published a special ‘F’ edition
dedicated to fiber optic. This issue covered every aspect of
fiber optics from cable manufacture to installation, applications and device technology. Ericsson’s transmission expertise
goes right back to the late 1920s when it began manufacturing loading coils, early signs of Ericsson’s ethos to provide the
telecoms industry with a wider range of products and services.
Flower power
and revolution
The cover of issue 3 in
1968 portrayed printed
circuit cards in the transfer
unit of Ericsson’s Stored
Program Controlled (SPC)
AKE exchange system. The
photograph is from the
automatic exchange
installed in Tumba (a suburb
of Stockholm). SPC
Ericsson Review,
exchanges were a milestone issue 3,
1968.
in the development of
telephony, as they made it easier to trace faults, and thus
reduce maintenace costs. The major concerns at the time
were capacity and cost of memory.
Oil and energy crises
Continued on
page 47...
The second issue of 1976
presented the AXE 10
switching system. The range of
articles in this issue highlighted
the shift in technology focus.
Telecoms was becoming much
more than wires, switches,
exchanges and transmission.
Hardware architecture and
design were now being
intimately combined with
software structure to provide
services, efficient traffic
Ericsson Review,
handling, management
issue 2, 1976.
systems and scalability.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
New network abstraction layers
20
Software-defined
networking: the service
provider perspective
An architecture based on SDN techniques gives operators greater freedom
to balance operational and business parameters, such as network resilience,
service performance and QoE against opex and capex.
AT T I L A TA K AC S , E L I S A BE L L AGA M BA , A N D JOE W I L K E
The traditional way of describing
network architecture and how a
network behaves is through the
fixed designs and behaviors of its
various elements. The concept
of software-defined networking
(SDN) describes networks and
how they behave in a more flexible
way – through software tools that
describe network elements in
terms of programmable network
states.
The concept is based on split architecture, which separates forwarding
functions from control functions. This
decoupling removes some of the complexity from network management,
providing operators with greater flexibility to make changes.
BOX A Ericsson’s approach to SDN goes beyond
the data center addressing issues in the
service-provider environment. In short
Ericsson’s approach is Service Provider
SDN. The concept aims to extend virtualization and OpenFlow – an emerging
protocol for communication between
the control and data planes in an SDN
architecture – with three additional key
enablers: integrated network control;
orchestrated network and cloud management; and service exposure.
There is no denying that networks
are becoming increasingly complex.
More and more functionality is being
integrated into each network element,
and more and more network elements
are needed to support evolving service
requirements – especially to support rising capacity needs, which are doubling
Terms and abbreviations
API
ARPU
CLI
DPI
GMPLS
L2
L3
L2-L4
M-MGW
MME
MSC-S
NAT
NMS
application programming interface
average revenue per user
command-line interface
deep packet inspection
generalized multi-protocol
label switching
Layer 2
Layer 3
Layers 2-4
Mobile Media Gateway
Mobility Management Entity
Mobile Switching Center Server
Network Address Translation
network management system
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
ONF
Open Networking Foundation
OSS/BSS operations and business support systems
PE
provider edge device
PGW
Packet Data Network Gateway
RG
residential gateway RWA
routing and wavelength
assignment
SDN
software-defined networking
SGW
Service Gateway
SLA
Service Level Agreement
VHG
virtual home gateway
VoIP
voice over IP
WAN
wide area network
every year1. One of the root causes of
network complexity lies in the traditional way technology has developed.
The design of network elements, such
as routers and switches, has traditionally been closed; they tend to have their
own management systems with vertically integrated forwarding and control
components, and often use proprietary
interfaces and features. The goal of
network management is, however, to
ensure that the entire network behaves
as desired – an objective that is much
more important than the capabilities
of any single network element. In fact,
implementing end-to-end networking
is an important mission for most operators, and having to configure individual network elements simply creates an
unwanted overhead.
Network-wide programmability – the
capability to change the behavior of the
network as a whole – greatly simplifies
the management of networks. And the
purpose of SDN is exactly that: to be able
to modify the behavior of entire networks in a controlled manner.
The tradition of slow innovation in
networking needs to be broken if networks are to meet the increased demand
for transport and processing capacity. By integrating recent technological
advances and introducing networkwide abstractions, SDN does just that.
It is an evolutionary step in networking.
Telephony has undergone similar
architectural transitions in the past.
One such evolution took place when a
clear separation between the functions
21
of the data plane (including SGW, PGW
and M-MGW) and the control plane
(including MME and MSC-S) was introduced. Now SDN has brought the concept of split architecture to networking.
As the business case proves, over the
next two to five years, SDN technology
will be deployed in networks worldwide.
At the same time, the need to maintain
traditional operational principles and
ensure interoperability between SDN
and more traditional networking components will remain. In the future,
SDN will help operators to manage
scale, reduce costs and create ­additional
­revenue streams.
Standardization
The goal of the Open Networking
Foundation (ONF), which was established in 2011, is to expedite the standardization of the key SDN interfaces.
Today, the work being conducted by
ONF focuses on the continued evolution of the OpenFlow protocol. ONF has
recently established the Architecture
and Framework Working Group, whose
goal is to specify the overall architecture of SDN. The work carried out by
this group will guide future standardization efforts based on strategic use cases, requirements for data centers and
carrier networks, the main interfaces, and their roles in the architecture.
Ericsson is actively driving the work of
this group, cooperating with other organizations to promote the evolution of
OpenFlow and support an open-source
implementation of the most recent
specifications.
Other standardization organizations,
most notably the IETF, have recently
begun to extend their specifications
to support SDN principles. In IETF, the
Interface to the Routing System (i2rs)
WG and the recent activity of the Path
Computation Element (PCE) WG will
result in standardized ways to improve
flexibility in changing how IP/MPLS networks behave. This is achieved through
the introduction of new interfaces to
distributed protocols running in the
network, and mechanisms to adapt network behavior dynamically to application requirements.
In addition to standardization organizations a multitude of active communities and open-source initiatives, such as
OpenStack, are getting involved in the
FIGURE 1 Service Provider SDN – components and promise
Service exposure
Northbound APIs to allow networks to respond
dynamically to application/service requirements
Orchestrated network and cloud management
Unified legacy and advanced network, cloud
management system and OSS/BSS to implement
SDN in step-by-step upgrade
Integrated network control
Control of entire network from radio to edge to
core to data center for superior performance
specification of various SDN tools, working on maturing the networking aspect
of virtualization.
Architectural vision
Split architecture – the decoupling of
control functions from the physical
devices they govern – is fundamental to
the concept of SDN. In split-architecture
networks, the process of forwarding in
the data plane is separated from the controller that governs forwarding in the
control plane. In this way, data-plane
and control-plane functions can be
designed, dimensioned and optimized
separately, allowing capabilities from
the underlying hardware to be exposed
independently.
This ability to separate control and
forwarding simplifies the development
and deployment of new mechanisms,
and network behavior becomes easier to manage, reprogram and extend.
Deploying a split architecture, however, does not remove the need for highavailability software and hardware
components, as networks continue to
meet stringent carrier requirements.
However, the decoupling approach to
architecture will help rationalize the
network, making it easier to introduce new functions and capabilities.
The ultimate goal of the SDN architecture is to allow services and applications to issue requests to the network
Service Provider
SDN
Service provider needs:
• Accelerated service
innovation
• Advanced public/hybrid
enterprise and consumer
cloud services
• Improved QoE
• Opex reduction through
centralized management
• Capex control
dynamically, avoiding or reducing the
need for human intervention to create
new services. This, compared to today’s
practices, will reduce the time to market
of new services and applications.
The OpenFlow protocol2 is supporting the separation of data and control
planes and allowing the path of packets through the network to be software
determined. This protocol provides a
simple abstraction view of networking
equipment to the control layer. Split
architecture makes virtualization of
networking resources easier, and the
control plane can provide virtual views
of the network for different applications
and tenants.
Ericsson has worked together with
service providers to understand their
needs for SDN both in terms of reducing costs and creating new revenue
opportunities. Based on these discussions, Ericsson has expanded the industry definition of SDN and customized it
to fit the needs of operators.
Ericsson’s hybrid approach – Service
Provider SDN – extends industry definitions including virtualization and
OpenFlow with three additional key
enablers:
integrated network control;
orchestrated network and cloud
management; and
service exposure.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
New network abstraction layers
22
FIGURE 2 Service Provider SDN – architectural vision
Ericsson NMS and cloud management system
Application
Application
Application
Tenants
SDN controller
Control
Control
(v) Control plane
APIs
(such as OpenFlow)
Forwarding
Forwarding
Forwarding
Forwarding
Integrated
systems
Routers
Physical
Virtual
Forwarding elements
Integrated network control
Service providers will use SDN across
the network from access, to edge to core
and all the way into the data center.
With integrated network control, operators can use their network features,
including QoS, edge functions and realtime activity indicators to deliver superior user experience.
By expanding the perspective of SDN
to include these three elements, service providers can evolve their existing network to the new architecture to
improve the experience of their customers. Implementing Service Provider SDN
should remove the dumb pipe label giving operators an advantage over competitors that do not own networks.
Orchestrated network
and cloud management
Service Provider SDN will integrate and
unify legacy network management systems with new control systems as well
as with OSS/BSS. The platform for integrated orchestration supports end-toend network solutions ranging from
access over aggregation to edge functions as well as the data centers used to
deliver telco and enterprise applications
and services.
Network virtualization
One of the benefits of Service Provider
SDN, especially from a network-­
spanning perspective is network virtualization. Through virtualization,
logical abstractions of a network can
be exposed instead of a direct representation of the physical network.
Virtualization allows logical topologies
to be created, as well as providing a way
to abstract hardware and software components from the underlying network
elements, thereby separating control
from forwarding capabilities and supporting the centralization of control.
Unified orchestration platforms support network programming at the highest layer as programming instructions
flow through the control hierarchy –
potentially all the way down to granular
changes in flow paths at the forwarding
plane level. Adding northbound APIs
Service exposure
Northbound APIs expose the orchestration platform to key network and
subscriber applications and services.
Together, the APIs and platforms allow
application developers to maximize
network capabilities without requiring
intimate knowledge of their topology
or functions.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Software
Hardware
into this unified orchestration layer provides the necessary support for applications and tenants to trigger automatic
changes in the network, ensuring optimal QoS and guaranteeing SLAs.
Unified and centralized orchestration platforms greatly simplify the process of configuring, provisioning and
managing complex service networks.
Instead of having to tweak hundreds of
distributed control nodes using fairly
complex CLI programming, operations
staff can use simple intuitive programming interfaces to quickly adjust network configurations and create new
services. By accelerating the process
of service innovation, Service Provider
SDN will lead to increased market share
and service ARPU, creating significant
revenue growth and possibly reducing
annual churn rates.
A high-level network architecture
that supports the Service Provider
SDN vision is illustrated in Figure   2.
Service-provider networks will combine
distributed control plane nodes (traditional routers and appliances), and dataplane elements that are governed by
centralized elements – SDN controllers
– in the control plane. Consequently,
to make service-provider networks programmable, distributed and centralized control-plane components must be
exposed to a unified orchestration platform. In addition, the key elements of
the orchestration platform and the control plane need to be exposed to network
and subscriber applications/services.
Application examples
The major use cases or applications of
SDN in service-provider networks are
summarized in Figure 3. Of these applications, the data center was the first to
make use of SDN. Ericsson’s approach to
this is described in several articles published in Ericsson Business Review3 and
in Ericsson Review4,5.
SDN can be applied in the aggregation network to support sophisticated
virtualization and to simplify the configuration and operation of this network segment. Ericsson has developed
a proof-of-concept system in cooperation with Tier 1 operators to evaluate the
applicability of SDN to aggregation networks. This work has been carried out
as part of the European Commission’s
Seventh Framework Programme6.
23
Ericsson and Telstra have jointly developed a service-chaining prototype that
leverages SDN technologies to enhance
granularity and dynamicity of service
creation. It also highlights how SDN
can simplify network provisioning and
improve resource utilization efficiency.
Packet-optical integration is a popular
topic of debate, out of which several different approaches are emerging. Split
architecture provides a simple way to
coordinate packet and optical networking; and so SDN, enhanced with features
such as routing and wavelength assignment (RWA) and optical impairments
management will be a natural fit for
packet-optical integration.
Ericsson has started to develop solutions to virtualize the home gateway.
Virtualization reduces the complexity
of the home gateway by moving most
of the sophisticated functions into the
network and, as a result, operators can
prolong the home gateway refreshment
cycle, cut maintenance costs and accelerate time to market for new services.
Virtualization of aggregation networks
The characteristics shared by aggregation and mobile-backhaul networks are
a large number of nodes and relatively static tunnels – that provide traffic
grooming for many flows. These networks are also known for their stringent
requirements with regard to reliability and short recovery times. Besides
L2 technologies IP and IP/MPLS is making an entrance as a generic backhaul
solution. From an operational point of
view, despite the availability of the distributed control plane technology, this
network segment is usually configured
statically through a centralized management system, with a point of touch
to every network element. This makes
the introduction of a centralized SDN
controller straightforward for backhaul
solutions.
A control element hosted on a telecomgrade server platform or on an edge
router provides the operator with an
interface that has the same look and
feel as a single traditional router. The
difference between operating an aggregation SDN network and a traditional
network lies in the number of touch
points required to provision and operate the domain. In the case of SDN,
only a few points are needed to control
the connectivity for the entire network. Consider, for example, an access/­
aggregation domain with hundreds or
even thousands of nodes running distributed IGP routing protocols and the
Label Distribution Protocol (LDP) to configure MPLS forwarding. In this case,
SDN principles can be applied to simplify and increase the scalability of provisioning and operating of such a network
by pulling together the configuration
of the whole network into just a few
control points.
The control element treats the underlying forwarding elements as remote
line cards of the same system and, more
specifically, controls their flow entries
through the OpenFlow protocol. With
this approach, any kind of connectivity model is feasible regardless of whether the forwarding node is L2 or L3 as,
from a pure forwarding point of view,
the same model is used in both flows.
At the same time, network resilience at
the transport level can be implemented by adding protection mechanisms
to the data path. The SDN controller
can pre-compute and pre-install backup routes and then protection switching is handled by the network elements
for fast failover. Alternatively, the SDN
controller can reroute around failures,
in case multiple failures occur, or in
scenarios that have less stringent recovery requirements.
From the outside, the entire network
segment appears to be one big PE router and, for this reason, neighboring network elements of the SDN-controlled
area cannot tell the difference (from a
protocol point of view) between it and a
traditional network. The network controller handles the interfacing process
with legacy systems for connection setup. Additional information on this point
can be found in a presentation on the
Virtual Network System7.
Dynamic service-chaining
For inline services, such as DPI, firewalls
(FWs), and Network Address Translation
(NAT), operators use different middleboxes or appliances to manage subscriber traffic. Inline services can be hosted
on dedicated physical hardware, or on
virtual machines. Service chaining is
required to route certain subscriber
traffic through more than one such service. There are still no protocols or tools
available for operators to perform flexible, dynamic traffic steering. Solutions
currently available are either static or
their flexibility is significantly limited
by scalability inefficiencies.
Given the rate of traffic growth, continued investment in capacity for
FIGURE 3 Application examples
Virtualization of
aggregation network
Network support
for cloud
Cloud/data
center
Mobile
Residental
Business
Virtual home
gateway
Policy-based
service chaining
Packet and optical
integration
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
New network abstraction layers
24
FIGURE 4 Service-chaining principles
FW
DPI
DPI
NAT
FW
SSR
SDN
switch
SDN
switch
Virtual Network System
domain
Flow before
SDN
switch
inline services needs to be managed
carefully. Dynamic service-chaining
can optimize the use of extensive hightouch services by selectively steering
traffic through specific services or
bypassing them completely which, in
turn, can result in capex savings owing
to the avoidance of over-dimensioning.
Greater control over traffic and
the use of subscriber-based selection
of inline services can lead to the creation of new offerings and new ways to
monetize networks. Dynamic service
steering enables operators to offer subscribers access to products such as virus
scanning, firewalls and content filters
through an automatic selection and subscribe portal.
This concept of dynamic service
chaining is built on SDN principles.
Ericsson’s proof-of-concept system uses
a logically centralized OpenFlow-based
controller to manage both switches and
middleboxes. As well as the traditional
5-tuple, service chains can be differentiated on subscriber behavior, application, and the required service. Service
paths are unidirectional; that is, different service paths can be specified for
upstream and downstream traffic.
Traffic steering has two phases. The
first classifies incoming packets and
assigns a service path to them based on
predefined policies. Packets are then
forwarded to the next service, based on
the current position in their assigned
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
service path. No repeating classification is required; hence the solution is
scalable.
The SDN controller sets up and reconfigures service chains flexibly to an
extent that is not possible with today’s
solutions. The dynamic reconfiguration
of service chains needs a mechanism to
handle notifications sent from middleboxes to the controller. For example, the
DPI engine notifies the controller that
it has recognized a video flow. These
notifications may be communicated
using the extensibility features of the
OpenFlow 1.x protocol.
Figure  4 summarizes service-­
chaining principles. The Virtual
Network System (VNS) is a domain of
the network where the control plane is
centralized which excludes some of the
traditional control agents. A simple API,
such as OpenFlow, can be used to control the forwarding functionality of the
network, and the VNS can create northbound interfaces and APIs to support
creation of new features, such as service
chaining, which allows traffic flows to
be steered dynamically through services or parts thereof by programming forwarding elements.
The services provided by the network may reside on devices located in
different parts of the network, as well
as within an edge router – for example,
on the service cards of Ericsson’s Smart
Services Router (SSR). Service chains are
Flow after
programmed into the network based on
a combination of information elements
from the different layers (L2-L4 and possibly higher). Based on operator policies,
various services can be applied to traffic
flows in the network.
For example, traffic may pass through
DPI and FW functions, as illustrated by
the red flow in Figure 4. However, once
the type of the flow has been determined by the DPI function, the operator may decide to modify the services
applied to it. For example, if the flow is
an internet video stream, it may no longer need to pass the FW service, reducing load on it. Furthermore, after the
service type has been detected, the subsequent packets of the same flow may
no longer need to pass the DPI service
either; hence the path of the flow can be
updated – as indicated by the blue flow
in Figure 4.
Packet-optical integration
The increased programmability that
SDN enables creates an opportunity to
address the challenges presented by
packet optical networking. SDN can
simplify multi-layer coordination and
optimize resource allocation at each layer by redirecting traffic (such as VoIP,
video and web) based on the specific
requirements of the traffic and the best
serving layer.
Instead of a layered set of separated
media coordinated in a static manner,
SDN could transform the packet-optical
infrastructure to be more fluid, with a
unified recovery approach and an allocation scheme based on real-time link
utilization and traffic composition.
The ONF still has some work to do to
adapt OpenFlow to cope with optical
constraints.
To speed up packet-optical integration, a hybrid architecture can be
deployed where OpenFlow drives the
packet domain, and the optical domain
remains under the control of GMPLS.
This approach utilizes the extensive
optical capabilities of GMPLS and,
therefore, instead working to extend
OpenFlow with optical capabilities, it
allows us to focus on the actual integration of optical and packet domains and
applications that utilize the flexibility of
a unified SDN controller.
25
Home gateway control
The concept of the virtual home gateway
(VHG) introduces a new home-network
architecture primarily driven by consideration to improve service delivery
and management. The target architecture emerges by applying SDN capabilities between the residential gateway
(RG) and the edge network – moving
most of the gateway’s functionalities
into an embedded execution environment. Virtualizing the RG significantly reduces its complexity and provides
the operator with greater granularity
in remote-control management, which
can be extended to every home device
and appliance. As a result, operators can
reduce their investments significantly
by prolonging the RG refreshment cycle,
cutting maintenance costs and accelerating time to market for new services.
The VHG concept allows operators to
offer seamless and secure remote profile
instantiation stretching the boundaries
of a home network without compromising security. The concept provides the
tools to configure and reconfigure middleboxes dynamically, so that communication between devices attached to
different home networks can be established, and/or provide specific connectivity requirements for a third-party
service provider – between, for example, a utility company and a particular
device. By embedding SDN capabilities,
Ericsson’s concept enables operators to
offer personalized applications to subscribers each with its own specific chain
of management policies and/or services.
The target architecture places an operator-controlled bridge at the customer’s premises instead of a complex
router, while the L3-L7 functionalities
are migrated to the IP edge or into the
operator cloud. Using SDN technology
between the IP edge and the switch in
this way offers the operator fine-grained
control for dynamic configuration of
the switch.
Conclusion
With its beginnings in data-center technology, SDN technology has developed
to the point where it can offer significant opportunities to service providers. To maximize the potential benefits
and deliver superior user experience,
SDN needs to be implemented outside
the sphere of the data center across the
entire network. This can be achieved
through enabling network programmability based on open APIs.
Service Provider SDN will help
operators to scale networks and take
advantage of new revenue-­generating
possibilities. The broader Service
Provider SDN vision goes beyond leveraging split architecture to include several software components that can be
combined to create a powerful end-toend orchestration platform for WANs
and distributed cloud data centers. Over
time, this comprehensive softwarebased orchestration platform will be
able to treat the overall operator network as a single programmable entity.
References
1. Ericsson, November 2012, Mobility Report, On the pulse of the Networked Society, available at: http://
www.ericsson.com/ericsson-mobility-report
2. Open Networking Foundation, available at: https://www.opennetworking.org/
3. Ericsson, November 2012, Ericsson Business Review, The premium cloud:
how operators can offer more, available at: http://www.ericsson.com/
news/121105-ebr-the-premium-cloud-how-operators-can-offer-more_244159017_c
4. Ericsson Review, December 2012, Deploying telecom-grade products in the cloud, available at: http://
www.ericsson.com/res/thecompany/docs/publications/ericsson_review/2012/er-telecom-gradecloud.pdf
5. Ericsson Review, December 2012, Enabling the network-embedded cloud, available at: http://www.
ericsson.com/res/thecompany/docs/publications/ericsson_review/2012/er-network-enabled-cloud.
pdf
6. European Commission, Seventh Framework Programme, Split Architecture Carrier Grade Networks,
available at: http://www.fp7-sparc.eu/
7. Elisa Bellagamba: Virtual Network System, MPLS & Ethernet World Congress, Paris, February 2012,
available at: http://www.slideshare.net/EricssonSlides/e-bellagamba-mewc12-pa8
Elisa Bellagamba
is a portfolio strategy
manager in Product Area
IP & Broadband, where
she has been leading SDN
related activities since their very
beginning. She holds an M.Sc. cum
laude in computer science engineering
from Pisa University, Italy.
Attila Takacs
is a research manager in
the Packet Technologies
Research Area of Ericsson
Research. He has been the
technical lead of research projects on
Software Defined Networks (SDN),
OpenFlow, GMPLS, Traffic Engineering,
PCE, IP/MPLS, Ethernet, and OAM for
transport networks. He is also an active
contributor to standardization; in
particular he has worked for ONF, IETF
and IEEE. He holds more than
30 international patent applications;
granted and in progress. He holds an
M.Sc. in computer science and a postgraduate degree in banking
informatics, both from the Budapest
University of Technology and
Economics, in Hungary. He has an MBA
from the CEU Business School,
in Budapest.
Joe Wilke
is head of Development
Unit IP & Broadband
Technology Aachen. He
currently leads the SDN
execution program and holds a degree
in electrical engineering from the
University of Aachen, Germany and a
degree in engineering and business
from the University of Hagen, Germany.
Acknowledgements
The authors would like to thank
Diego Caviglia, Andreas Fasbender,
Howard Green, Wassim Haddad,
Alvaro de Jodra, Ignacio Más,
Don McCullough and
Catherine Truchan for their
contributions to this article.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Smarter networks
26
HSPA evolution: for future mobilebroadband needs
As HSPA continues to evolve, addressing the needs of changing user behavior, new techniques develop
and become standardized. These techniques provide network operators with the flexibility, capacity and
coverage needed to carry voice and data into the future.
N I K L A S JOH A N S S ON, L I N DA BRU S , E R I K L A R S S ON, BI L LY HO GA N A N D P E T E R VON W RYC Z A
Mobile broadband (MBB),
providing high-speed internet
access from more or less
anywhere, is becoming a reality
for an increasing proportion of
the world’s population. There
are several factors fuelling the
need for high-performance
MBB networks, not the least,
the growing number of mobile
internet connections. As Figure 1
illustrates, global mobile
subscriptions (excluding M2M)
are predicted to grow to 9.1 billion
by the end of 2018. Nearly 80
percent of mobile subscriptions
will be MBB ones1 , indicating that
MBB will be the primary service
for most operators in the coming
years.
Impact of affordable smartphones
To a large extent, the rapid growth of
MBB can be attributed to the widespread
availability of low-cost MBB-capable
smartphones, which are replacing
BOX A voice-centric feature phones. For less
than USD 100, consumers can purchase
highly capable WCDMA/HSPA-enabled
smartphones with dual-core processors
and dual-band operation that support
data rates up to 14.4Mbps. This priceto-­sophistication ratio has turned the
smartphone into an affordable massmarket product, and has accelerated the
increase in smartphone subscriptions –
­estimated to rise from 1.2 billion at the
end of 2012 to 4.5 billion by 20181.
Ericsson ConsumerLab studied a
group of people to assess how they perceived network quality and what issues
they encountered when using their
smartphones. The study identified two
key factors that are essential to the perceived value of a smartphone: a fast and
reliable connection to the data network,
and good coverage2.
These findings highlight an important goal for operators: to provide all
network users with high-speed data services and good-quality voice ­services
everywhere. This can be achieved by
securing:
Terms and abbreviations
CELL_FACH Cell forward access channel
CPC
Continuous Packet Connectivity
DPCH Dedicated Physical Channel
EULEnhanced Uplink
HS-DSCH High-Speed Downlink Shared Channel
HSDPA
High-speed Downlink
Packet Access
HSPA
High-speed Packet Access
HSUPA
High-speed Uplink Packet Access
LPN
low-power node
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
M2Mmachine-to-machine
MBB
mobile broadband
MIMO
multiple-input multiple-output
ROT
rise-over-thermal
SRB
Signaling Radio Bearer
ULuplink
URA_PCH UTRAN registration area
paging channel
UTRAN
Universal Terrestrial
Radio Access Network
WCDMA
Wideband Code Division Multiple Access
capacity – to handle growing
smartphone traffic cost-efficiently;
flexibility – to manage the wide range of
traffic patterns efficiently; and
coverage – to ensure good voice and app
user experience everywhere.
App coverage
For smartphone applications, like social
networking and video streaming, to
function correctly, access to the data network and a network that can deliver a
defined minimum level of performance
is needed. The relationship between the
performance requirements (in terms
of data speed and response time) of an
application and the actual performance
delivered by the network for that user
at their location at a given time determines how well the user perceives the
performance of the application.
The term app coverage denotes the
level of network performance needed to
provide subscribers with a satisfactory
user experience for a given application.
In the past, the task of dimensioning networks was simpler, as calculations were based on delivering target
levels of voice coverage and providing
a minimum data rate. Today’s applications, however, have widely varying performance requirements. As a
result, dimensioning a network has
become a more dynamic process and
one that needs to take these varying
performance requirements into consideration, for apps that are currently
popular with subscribers.
Footprint
Illustrated in Figure 2, at the end of
2012, 55 percent of the world’s population was covered by WCDMA/HSPA,
a figure that is set to rise to 85 percent
27
Evolution of traffic patterns
Applications have varying demands and
behaviors when it comes to when and
how much data they transmit. Some
apps transmit a large amount of data
continuously for substantial periods
of time and some transmit small packets at intervals that can range from a
few seconds to minutes or even longer.
Applications have varying demands,
typically sending lots of data in bursts,
interspersed with periods of inactivity
when they send little or no data at all.
Rapid handling of individual user
requests, enabled by high instantaneous
data rates, improves overall network
performance as control-channel overhead is reduced and capacity for other
traffic becomes available sooner. So, if
a network can fulfill requests speedily,
all users will experience the benefits of
reduced latency and faster round-trip
times.
Web browsing on a smartphone is a
classic example of a bursty application,
both for uplink and downlink communication. When a smartphone requests
the components of a web page from the
network (in the uplink) they are transferred in bursts (in the downlink), and
the device acknowledges receipt of
the content (in the uplink). As a result,
FIGURE 1 Mobile and MBB subscriptions (2009-2018)1
Subscriptions/lines (million)
Mobile subscriptions
Mobile broadband
10,000
9,000
8,000
7,000
6,000
5,000
4,000
3,000
2,000
1,000
0
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
uplink and downlink performance
becomes tightly connected and therefore better uplink performance has a
positive effect on downlink data rates
as well as overall system throughput.
For web browsing, the instantaneous
downlink speed for mobile users needs
to be much higher on average than the
uplink speed. However, the number of
services requiring higher data rates in
FIGURE 2 the uplink, such as video calling and
cloud synching of smartphone data, is
on the rise.
As user behavior changes, traffic-­
volume patterns also change, and measurements show it is becoming more
common for uplink levels to be on par
with downlink levels, and in some
­cases even outweigh the downlink traffic. Consequently, continuing to
Population coverage by technology (2012-2018)
(Source: Ericsson1)
100
>85%
>90%
>85%
80
% population coverage
by the end of 20181. Today, many developed markets are nearing the 100 percent population coverage mark3. This
widespread deployment, together with
support for the broadest range of devices, makes WCDMA/HSPA the primary
radio-access technology to handle the
bulk of MBB and smartphone traffic for
years to come.
Since its initial release, the 3GPP
WCDMA standard has evolved, and
continues to develop. Today, WCDMA/
HSPA is a best-in-class voice solution
with exceptional voice accessibility and
retainability. It offers high call retention
as well as being an excellent access technology for MBB, as it delivers high data
rates and high cell-edge throughput – all
of which enable good user experience
across the entire network.
The continued evolution of WCDMA/
HSPA in Releases 11 and 12 includes several key features that aim to increase
network flexibility and capacity to meet
growing smartphone traffic and secure
voice and app coverage.
~60%
60
~55%
40
20
0
~10%
2012
2018
GSM/EDGE
2012
2018
WCDMA/HSPA
2012
2018
LTE
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Smarter networks
28
FIGURE 3 develop data rates to secure uplinkheavy services is key to improving overall user performance.
Where to improve, densify and add
Area traffic density
Improve
Densify
Add
Improve
Densify
Improve
Dense
urban
FIGURE 4 Urban
Suburban
Rural
Relationship between maximum interference and peak rate
UL ROT
Legend
Y=
Y
X=
X
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Rate
Maximum interference
handled by the network
Maximum uplink data
rate that can be achieved
High performance networks
The standard approach used to create a
high-performance network with wide
coverage and high capacity is to first
improve the macro layer, then ­densify it
by deploying additional macro base stations, and finally add low power nodes
(LPNs) in strategic places, such as traffic hotspots, that can offload the ­macro
network.
Each step addresses specific performance targets and applies to different
population densities, from urban to
rural – as illustrated in Figure 3. The
evolution of WCDMA/HSPA includes a
number of features that target ­macro
layer improvement and how deployments where LPNs have been added can
be enhanced.
Improving the uplink
Features in the 3GPP specification have
recently achieved substantial improvement of uplink capabilities. Features
such as uplink multi-carrier, higherorder modulation with MIMO, EUL
in CELL_FACH state, and Continuous
Packet Connectivity (CPC) have multiplied the peak rate (up to 34Mbps per
carrier in Release 11) and increased the
number of simultaneous users a network can support almost fivefold.
Given the high uplink capabilities
already supported by the standard,
the next development (Release 12) will
enable and extend the use of these capabilities to as many network users as
possible.
The maximum allowed uplink interference level in a cell, also known as
maximum rise-over-thermal (ROT), is a
highly important quantifier in WCDMA
networks. This is because the maximum allowed interference level has a
direct impact on the peak data rates that
the cell can deliver.
Typically, macro cells are dimensioned with an average ROT of around
7dB, which enables UL data rates of
5.7Mbps (supported by most commercial smartphones), and secures voice
and data coverage for cell-edge users.
High data rates, such as 11Mbps (available since release 7) and 34Mbps (available since release 11) require ROT levels
29
greater than 10dB and 20dB respectively. Figure 4 illustrates the relationship
between ROT and peak data rate.
The maximum uplink interference
level permissible is determined by a
number of factors including the density
of the network, the capability of the network to handle interference (for example with advanced techniques such as
Interference Suppression), and the capabilities of the devices in the network,
including both smartphones and ­legacy
feature phones.
The Lean Carrier solution, introduced
in Release 12, is an additional ­capability
that helps operators meet the needs of
high-data-rate users. This ­multi-carrier
solution is built on the Release 9 HSUPA
dual-carrier one that is currently being
implemented in commercial smartphones. The dual-carrier solution allows
two carriers, primary and secondary,
to be assigned to a user. By doing this,
the traffic generated by the user can
be allocated in a flexible way between
the two carriers, while at the same
time doubling the maximum peak rate
achievable.
The Lean Carrier solution optimizes
the secondary carrier for fast and flexible handling of multiple high-data-rate
users, through more efficient granting
and lower cost per bit. The solution is
designed to support multiple bursty
data users in a cell transmitting at the
highest peak rates without causing any
uplink interference among themselves
or to legacy users. To maximize energy efficiency, the Lean Carrier solution
should cost nothing in system or terminal resources on the secondary carrier
until the user starts to send data.
Lean Carrier can be flexibly deployed
according to the needs of the network.
For example, the maximum ROT on a
user’s secondary (lean) carrier can be
configured to support any available
uplink peak data rate, while the maximum ROT on a user’s primary carrier
can be configured to secure cell-edge
coverage for signaling, random access
and legacy (voice) users.
Rate adaptation is another technology under study that results in increased
network capacity for some common
traffic scenarios, such as areas where
subscribers are a mix of high and lowrate users or areas where there are only
high-rate users. High uplink data rates
FIGURE 5 Rate adaptation results in predictable interference levels
Received power
Baseline: Fixed rate
variable power
Rate adaptation: Fixed
received power and
variable rate
DATA
DATA
Control
Control
Time
require more power. Maintaining a
fixed data rate at the desired ­quality
target in an environment where
interference levels vary greatly
can result in large fluctuations in
received power. To avoid such fluctuations, the concept of rate adaptation can be applied. High-rate users
are assigned with a fixed receivedpower budget, and as interference
levels change, bit rates are adapted
to maintain the desired quality target, while not exceeding the allowed
power budget. In short, as illustrated
in Figure 5, the bit rate is adapted to
received power, and not the power
to the rate.
Limiting fluctuations in received
power for high-rate users is good
for overall system capacity because
these high-rate users can transmit
more efficiently, and other users in
the system, including low-rate ones
such as voice users, consume less
power when power levels are stable
and predictable.
Maintaining a device in connected
mode for as long as possible is another technique that can be used to
improve performance of the uplink.
Smartphone users want to be able
to rapidly access the network from
a state of inactivity. Maintaining a
device in a connected-mode state,
such as CELL_FACH or URA_PCH,
for as long as possible is one way of
achieving this – access to the network
from these states is much faster than
from the IDLE state. In recent releases,
connected mode has been made more
efficient from a battery and resource
point of view through the introduction
of features such as CPC, fractional DPCH
and SRB on HS-DSCH. As a consequence
it is now feasible to maintain inactive
devices in these states for longer.
As the number of smartphone
users increases, networks need flexible mechanisms to maintain high system throughput, even during periods
of extremely heavy load. Allowing the
network to control the number of concurrently active users, as well as the
number of random accesses, is one such
mechanism.
Improvements that enable high
throughput under heavy load, and allow
users to benefit from lower latency in
connected mode, while enabling service-differentiated admission decisions
and control over the number of simultaneous users, have been proposed for
Release 12.
Expanding voice and app coverage
Good coverage is crucial for positive
smartphone user experience and customer loyalty2, which for operators
translates into securing voice coverage and delivering data-service coverage that meets the needs of current and
future apps.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Smarter networks
30
FIGURE 6 Release 11 uplink transmit diversity beamforming
BOX B  
The system
The scenario
shown in
Figure 7 is for
bursty traffic.
Four LPNs have
been added
to each macro
base station in
the network,
and 50 percent
of the users are
located in traffic
hotspots. The
transmission
power for the
macro base
station was
20W, and 1W and
5W LPNs were
deployed.
There are several ways to improve
coverage for voice and data. One way is
to use lower frequency bands – when
compared to 2GHz bands, considerable
coverage improvement can be achieved
by refarming the 900MHz spectrum
FIGURE 7 from GSM, for example. Voice coverage
can be significantly extended with
­lower-rate speech codecs, whereas fourway receiver diversity and advanced
antennas can improve coverage for both
voice and data.
System-level gains – for scenario described in Box B
User throughput gain (percent)
300
1W
5W
250
200
150
100
50
0
Average
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Cell edge
LPNs were
deployed
randomly and
no LPN range
expansion was
used. Gains are
given relative
to a macro-only
deployment.
Offloading was
32 percent for
1W LPNs and 41
percent for 5W
LPNs, where
offloading is a
measure of the
percentage of
traffic served by
the LPN.
Uplink transmit diversity was introduced in Release 11. This feature supports terminals with two antennas to
increase the reliability and coverage
of uplink transmissions and decrease
overall interference in the system. It
works by allowing the device to use both
antennas for transmission in an efficient way using beamforming. Figure 6
illustrates how the radio transmission
becomes focused in a given direction,
resulting in a reduction in interference between the device and other
nodes, and improving overall ­system
performance.
An additional mode within uplink
transmit diversity is antenna selection.
Here, the antenna with the best radio
propagation conditions is chosen for
transmission. This is useful, for example, when one antenna is obstructed
by the user’s hand. Uplink transmit
diversity increases the coverage of all
uplink traffic for voice calls and data
transmissions.
With Release 11, multi-flow HSDPA
transmissions are supported. This
allows two separate nodes to transmit
to the same terminal, improving performance for users at the cell edge and
resulting in better app coverage.
In Release 12, simultaneous app
data and voice call transmissions will
become more efficient, and the time it
takes to switch the transmission time
interval from 10ms to 2ms is considerably shorter. These improvements
increase both voice and app coverage.
Enhancing small-cell deployments
The addition of small cells through
deploying LPNs in a macro network –
resulting in a heterogeneous network
– is a strategic way to improve capacity,
data rates and coverage in urban areas.
Typically, the deployment of LPNs is beneficial in hotspots where data usage is
heavy, to bridge coverage holes created
by complex radio environments, and in
some specific deployments such as inbuilding solutions.
Figure 7 shows the performance
gains in a heterogeneous-network
deployment (which is described in more
detail in Box B). Offloading to small
cells has several benefits: it provides
increased capacity for handling smartphone traffic, and it results in enhanced
app coverage.
31
To maximize spectrum usage, the traditional macro base stations and LPNs
share the same frequency, either with
separate or shared cell identities. These
deployments, illustrated in Figure 8,
are referred to as separate cell and combined cell.
It is possible to operate both separate
and combined-cell deployments based
on functionality already implemented
in the 3GPP standard, and such deployments have been shown to provide substantial performance benefits over
macro-only deployments.
Today, combined cells tend to be
deployed in specific scenarios, such as
railroad, highway and in-building environments. Separate-cell deployments,
on the other hand, are more generic
and provide a capacity increase in more
common scenarios.
In 3GPP Release 12, small-cell range
expansion techniques and control channel improvements are being introduced
to enable further offloading of the
­macro network. Mobility performance
enhancements for users moving at high
speeds through small cell deployments
are also being investigated by 3GPP.
When a macro cell in a combinedcell deployment is complemented with
additional LPNs close to users, the data
rate and network capacity is improved.
By allowing the network to reuse the
same spreading codes in different parts
of the combined cell, the cell’s capacity can be further increased – a technique being studied in Release 12. And
as there is no fundamental uplink/
downlink imbalance in a combined cell,
mobility signaling is robust, signaling
load is reduced, and network management is simplified.
In summary, heterogeneous networks are essential for handling growing smartphone traffic because they
support flexible deployment strategies,
increase the capacity of a given HSPA
network, and extend voice and app coverage. The improvements standardized
in Release 12 will further enhance these
properties.
Conclusions
WCDMA/HSPA will be the main technology providing MBB for many years
to come. Operators want WCDMA/HSPA
networks that can guarantee excellent
user experience throughout the whole
FIGURE 8 LPN deployment scenarios
LPN
Macro
LPN
LPN
LPN
Macro
LPN
LPN
RNC
LPNs deployed as separate cells
on the same carrier
RNC
LPNs deployed as part of a combined cell
on the same carrier
network coverage area for all types of
current and future mobile devices. The
prerequisites for networks are:
capacity – to handle growing
smartphone traffic cost-efficiently;
flexibility – to manage the wide range of
traffic patterns efficiently; and
coverage – to ensure good voice and app
user experience everywhere.
HSPA evolution, through the capabilities already available in 3GPP and those
under study in 3GPP Release 12, aims
to fulfill these prerequisites. There are
several ways to improve voice and app
coverage. Enhancements to the uplink
improve the ability to quickly and efficiently serve bursty traffic – improving user experience and increasing
smartphone capacity. Small-cell
improvements will increase network
capacity for smartphone traffic and further improve voice and app coverage.
With all of these enhancements,
WCDMA/HSPA, already the dominant
MBB and best-in-class voice ­technology,
has a strong evolution path to meet
the future demands presented by the
growth of MBB and highly capable
smartphones globally.
References
1. Ericsson Mobility Report, June 2013, available at:
http://www.ericsson.com/res/docs/2013/ericsson-mobility-report-june-2013.pdf
2. Ericsson ConsumerLab report, January 2013, Smartphone usage experience –
the importance of network quality and its impact on user satisfaction, available
at: http://www.ericsson.com/news/130115-ericsson-consumerlab-reportnetwork-quality-is-central-to-positive-smartphone-user-experiences-andcustomer-loyalty_244129229_c
3. International Communications Market Report 2011, Ofcom, available at: http://
stakeholders.ofcom.org.uk/binaries/research/cmr/cmr11/icmr/ICMR2011.pdf
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Smarter networks
32
Niklas Johansson
Peter von Wrycza
is a senior researcher at
Ericsson Research. He
joined Ericsson after
receiving his M.Sc. in
engineering physics and B.Sc. in
business studies from Uppsala
University in 2008. Since joining
Ericsson, he has been involved in
developing advanced receiver
algorithms and multi-antenna
transmission concepts. Currently, he is
project manager for the Ericsson
Research project that is developing
concepts and features for 3GPP
Release 12.
is a senior researcher at
Ericsson Research, where
he works with the
development and
standardization of HSPA. He received
an M.Sc. (summa cum laude) in
electrical engineering from the Royal
Institute of Technology (KTH),
Stockholm, Sweden, in 2005, and was
an electrical engineering graduate
student at Stanford University,
Stanford, CA, in 2003-2005. In 2010,
he received a Ph.D. in
telecommunications from KTH.
Billy Hogan
Erik Larsson
joined Ericsson in 1995
and works in the Technical
Management group in the
Product Development
Unit WCDMA and MultiStandard RAN. He is a senior specialist
in the area of enhanced uplink for
HSPA. He works with the system design
and performance of EUL features and
algorithms in the RAN product, and
with the strategic evolution of EUL to
meet future needs. He is currently team
leader of the EUL Enhancements team
for 3GPP release 12. He holds a B.E. in
electronic engineering from the
National University of Ireland, Galway,
and an M.Eng in electronic engineering
from Dublin City University, Ireland.
joined Ericsson in 2005.
Since then has held
various positions at
Ericsson Research,
working with baseband algorithm
design and concept development for
HSPA. Today, he is a system engineer in
the Technical Management group in the
Product Development Unit WCDMA
and Multi-Standard RAN and works
with concept development and
standardization of HSPA. He holds an
M.Sc. in engineering physics (1999)
and a Ph.D. in signal processing (2004),
both from Uppsala University, Sweden.
Linda Brus
joined Ericsson in 2008.
Since then, she has been
working with system
simulations, performance
evaluations, and developing algorithms
for WCDMA RAN. Today, she is a
system engineer in the Technical
Management group in the Product
Development Unit WCDMA and MultiStandard RAN, working with concept
development for the RAN product and
HSPA evolution. She holds a Ph.D. in
electrical engineering, specializing in
automatic control (2008) from
Uppsala University, Sweden.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Same bandwidth, double the data
33
Next generation
video compression
MPEG and ITU have recently approved a new video-compression
standard known as High Efficiency Video Coding (HEVC), or H.265, that
is set to provide double the capacity of today’s leading standards1.
P E R F RÖJ DH , A N DR E Y NOR K I N A N D R IC K A R D S JÖBE RG
Requiring only half the bitrate
of its predecessor, the new
standard will significantly reduce
the need for bandwidth and
expensive, limited spectrum.
HEVC will enable new video
services to be launched, and the
first applications that are likely to
appear will be for mobile devices
and OTT applications, followed
by TV – and in particular ultra HD
television (UHDTV).
State-of-the-art video compression can
reduce the size of raw video by a factor of about 100 without any noticeable
reduction in visual quality. Estimates
indicate that compressed real-time video accounts for more than 50 percent of
current network traffic2, and this figure is set to rise to 90 percent within a
few years3.
New services, devices and changing
viewing patterns are among the factors contributing to this growth, as is
increased viewing of traditional TV
and video-streaming services, such as
Netflix, YouTube and Hulu, on a range of
BOX A devices – from phones and tablets to PCs
and home-entertainment systems. As
HD shifts from luxury to commodity, it
will soon be challenged by UHD, which
offers resolutions up to 16 times greater.
Making standards
Most video viewed by subscribers today
has been digitized and reduced in
size through the application of a compression standard. The more popular
include the H.26x series from ITU and
the MPEG-x series from ISO/IEC. First
published in 1994, the MPEG-2 standard, also known as H.262, played a
crucial role in the launch of digital-TV
services as it enabled the compression
of TV streams to fit the spectrum available. This is also the standard used to
compress movies onto a DVD.
The H.264 standard (also known as
MPEG-4 AVC), published in 2003, has
provided the best compression efficiency to date, and is currently the most
widely used video-compression codec. It
has been successfully incorporated into
most mobile devices, and is the best way
to reduce the size of video carried over
Terms and abbreviations
AVC
advanced video coding
CABAC
context-adaptive binary
arithmetic coder
CTU
coding-tree unit
CU
coding unit
fps
frames per second
HD
high definition; often refers to
1280 x 720 or 1920 x 1080 pixels
HEVC
High Efficiency Video Coding
IEC
International Electrotechnical Commission
ISO
International Organization for Standardization
ITU
International Telecommunication Union
MPEG
Moving Picture Experts Group
OTTover-the-top
SAO
sample adaptive offset
UHD
ultra high definition: often refers to
3840 x 2160 (4K) or
7680 x 4320 (8K) pixels
WPP
wavefront parallel processing
the internet. It is the preferred format
for Blu-ray discs, telepresence streams
and, most notably, HDTV.
Now imagine a codec that is twice
as efficient as H.264. This was the target set by MPEG and ITU in 2010, when
they embarked on a joint standardization effort that three years later delivered HEVC/H.2654,5.
The new codec offers a much more
efficient level of compression than
its predecessor H.264, and is particularly suited to higher-resolution video
streams, where bandwidth savings with
HEVC are around 50 percent. In simple
terms, HEVC enables a network to deliver twice the number of TV channels.
Compared with MPEG-2, HEVC can provide up to four times the capacity on the
same network.
Like most standards, the MPEG and
ITU video codecs have been developed
in a collaborative fashion involving
many stakeholders – manufacturers,
operators, broadcasters, vendors and
academics. Ericsson has been an active
participant in video standardization
for more than 15 years, and was closely
involved in HEVC.
Throughout the development of the
standard, Ericsson has led several of
the core experiments, chaired ad-hoc
working groups and contributed significantly to the development of the technology behind the codec. Our greatest
expertise lies in the areas of the deblocking filter6 and in reference picture
management7.
Concepts that create efficiency
One of the primary target areas for
HEVC compression is high-­resolution
video, such as HD and UHD. The statistical characteristics of these types
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Same bandwidth, double the data
34
FIGURE 1 Simplified HEVC encoder diagram
Input signal
T
Transform coefficients
Q
Q-1
T-1
Entropy
coding
In-loop
filters
Intraprediction
Motion
compensation
Decoded
picture buffer
Motion vectors
Motion
estimation
of video streams tend to be different
from lower-­resolution content: frame
sizes are larger, and frame rates and perceived quality are higher – imposing
tough requirements on compression
efficiency, as well as on the computational complexity of the encoding and
decoding processes.
As the architectures of smartphone
s and tablets go multi-core, the ability
to take advantage of parallel processing is key when it comes to the efficient
compression of high-resolution content.
All of these points have been taken into
FIGURE 2 consideration during the development
of the new standard.
The hybrid block-based coding used
by the new codec is the same as the
one used in earlier video-coding standards. To encode content, video frames
are divided into blocks that are coded
individually by applying prediction –
based either on neighboring blocks in
the same picture (intra prediction) or
from previously coded pictures (motion
estimation/compensation).
The difference between the predicted result and original video data is
Example of the coding-tree unit structure in HEVC
CTU
CTU structure
CU
CU
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
CU
subsequently coded by applying block
transforms and quantization. In this
way, a block can be represented by just
a few non-zero coefficients. Quantized
transform coefficients, motion vectors,
prediction directions, block modes and
other types of information are encoded with lossless entropy coding. Hybrid
block-based coding is illustrated in
Figure  1.
To ensure the highest level of compression efficiency, and support for parallel processing, some parts of HEVC
have been significantly modified compared with previous generations of
hybrid block-based codecs. For most of
the previous MPEG-x and H.26x codecs,
the largest entity that could be independently encoded was a macroblock
(16 × 16 pixels). For HEVC, the picture
is split into coding-tree units (CTUs)
with a maximum size of 64 × 64 pixels. Every CTU is the root of a quadtree,
which can be further divided into leaflevel coding units (CUs), as illustrated
in Figure  2. The CTUs are coded in raster scan order, and each unit can itself
contain a quadtree structure. Each CU
contains one or more prediction partitions that are predicted independently of each other. A CU is also associated
with a transform quadtree that compresses the prediction residual and has
a structure similar to that of a CTU – as
shown in Figure  2.
Partitions for motion prediction can
form square or rectangular shapes,
which is also the case with earlier standards. HEVC also supports something
called asymmetric motion partitioning, which can split the CU into prediction units of unequal width or height, as
illustrated in Figure  3.
The size of the prediction blocks in
HEVC can therefore vary from 4 × 4 samples up to 64 × 64, while transform sizes
vary from 4 × 4 to 32 × 32 samples. Large
prediction blocks and transform sizes
are the most efficient way to encode
large smooth areas, whereas smaller
prediction blocks and transforms can
be used to achieve precision in areas
that contain finer detail.
The HEVC specification covers more
intra-prediction modes than H.264,
including a planar mode to approximate a surface from neighboring
pixels, a flat mode and 33 angular prediction modes. Motion-compensated
35
prediction for luma transform blocks
is performed with up to quarter-pixel
precision, whereas motion compensation for color components is performed
with one-eighth-of-a-pixel precision.
Interpolation for fractional pixel positions uses 8-tap filters for luma blocks
and 4-tap filters for color.
In HEVC there is a single entropy
coder for low-level data. This is the context-adaptive binary arithmetic coder
(CABAC), which is similar to the one
used in H.264, but modified to facilitate
parallel processing. Higher-level information, such as sequence parameters,
is encoded with variable-length or fixedlength encoding.
HEVC defines two in-loop filters: a
deblocking filter and a sample adaptive
offset (SAO) filter. The latter is applied to
the output of the deblocking filter, and
increases the quality of reference pictures by applying transmitted offsets
to samples that fulfill certain criteria.
In-loop filters improve the subjective
quality of reconstructed video as well
as compression efficiency. Deblocking
filtering in HEVC is less complex than
that of H.264, as it is constrained to an
8 × 8 block grid. This constraint, together with filtering decisions and operations that are non-overlapping between
two boundaries, simplifies multi-core
processing.
Parallel processing
To make the most of the increasingly
widespread use of multi-core processors, plus the ever-growing number of
cores used in consumer-class processors, significant attention was paid to
the parallelization characteristics of
­video encoding and decoding when
designing HEVC. As it is computationally more complex than its predecessor,
maximizing parallelization has been
a key factor in making HEVC an efficient real-time encoding and decoding
solution.
Several HEVC tools have been
designed for easy parallelization. The
deblocking filter can be applied to 8 × 8
pixel blocks separately, and transformcoefficient-coding contexts for several
coefficient positions can be processed
in parallel. Tiles and wavefront parallel processing (WPP) are among several
HEVC tools that can provide high-level
parallelism.
FIGURE 3 Possible motion prediction partitions in HEVC
Asymmetric motion partitions are shown in the bottom row.
Only square partitions are allowed for intra prediction
The concept behind WPP is to re-initialize CABAC at the beginning of each
line of CTUs.
To facilitate CABAC adaptation to the
content of the video frame, the coder is
initialized once the statistics from the
decoding of the second CTU in the previous row are available.
Re-initialization of the coder at the
start of each row makes it possible to
begin decoding a row before the processing of the preceding row has been
completed. Thus, as shown in the example in Figure  4, several rows can be
decoded in parallel in several threads
with a delay of two CTUs between two
consecutive rows.
The Tiles tool can be used for parallel encoding and decoding, and works
by dividing a picture into rectangular
areas (tiles) – as shown in Figure  5 –
where each tile consists of an integer
number of CTUs. The CTUs are processed in a raster scan order within
each tile, and the tiles themselves are
processed in the same way. Prediction
based on neighboring tiles is disabled,
and so the processing of each tile is independent. In-loop filters, however, can
operate over tile boundaries. And
FIGURE 4 Multi-thread decoding with wavefronts. Gray areas
indicate CTUs that have already been decoded
CTU
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Same bandwidth, double the data
36
FIGURE 5 Example of the way an image can be divided into tiles
Column boundaries
CTU
CTU
tile 1
tile 2
tile 3
tile 4
tile 5
tile 6
tile 7
tile 8
tile 9
Row boundaries
CTU
as deblocking and SAO can be parallelized, filtering can be performed
­independently inside each tile, and tile
boundaries can be processed by in-loop
filters in a final pass.
The HEVC standard therefore enables
both high- and low-level parallelization,
which can provide significant benefits
for multi-thread encoding and decoding of video such as 4K and 8K that has a
higher resolution than HD.
Performance and complexity
The improved coding efficiency of
HEVC does however come with a price
tag: increased computational complexity. Compared with its predecessor,
HEVC is 50 -100 percent more complex
for decoding and up to 400 percent more
complex when it comes to encoding.
While these comparisons are based on
preliminary tests, they do give an indication of the new codec’s computational complexity.
Real-time implementations of HEVC
demonstrate that decoding of full HD
(1080p) at 50 or 60fps is possible on fast
desktop and laptop computers, running
on a single core. Performance increases with multiple core implementations
(hardware acceleration), so that a modern smartphone is capable of 1080p
decoding at 25 or 30fps8.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Applications
The new standard is a general one suitable for the compression of all kinds
of video. The focus for the first version
is consumer applications and for this,
three profiles have been defined: Main,
Main 10 and Main Still Picture.
Main is an all-purpose profile with
a depth of 8 bits per pixel, supporting
4:2:0 – the most common uncompressed
video format used by consumer devices
from mobile phones to HDTVs. Main 10
extends the bit depth to 10 bits per pixel,
which is well suited to consumer applications, such as UHDTV, where very
high quality is critical. The increased
bit depth can compress wide ­dynamic
range video without creating banding artifacts, which sometimes occurs
with 8 bits. The third profile, Main Still
Picture, used for still images, is a subset
of Main and carries a single still picture
at a depth of 8 bits per pixel.
The initial deployments of HEVC
released in 2013 will be for mobiles
and OTT applications. Software implementations capable of decoding HEVC
without hardware acceleration can
easily be downloaded to smartphones,
tablets and PCs, enabling mobile TV,
streaming and download services on
existing devices. To this end, in August
2012, Ericsson announced SVP 55009,
the world’s first HEVC real-time video
encoder for live-TV delivery to mobile
devices. However, as it is better to perform encoding on hardware and as
HEVC is computationally more demanding than previous standards, it may
be some time before video telephony
based on this standard enters mobile
platforms, whereas encoding on PCs is
already feasible.
Set-top boxes with new decoders will
become available soon, enabling content broadcast via satellite, cable or terrestrially to take advantage of HEVC.
The new standard plays a key role in
the provision of UHDTV, and as prices
drop and displays become affordable,
the number of services utilizing such
high resolutions is expected to rise
within a few years. Flat-panel displays
for HDTV have been on the market for
almost a decade, so this may be a good
time for consumers to start upgrading
to UHDTV.
What’s coming
The finalized version of HEVC targets
most consumer devices and services.
However, for more specialized applications, such as 3D, content production or
heterogeneous devices and networks,
some additions to HEVC may prove useful. With this in mind, MPEG and ITU
are working together on a number of
ideas, including support for stereo and
multi-view (glasses-free) 3D video, an
extension that encodes multiple views
by rearranging picture buffers and
reference picture lists. A first drop is
expected in January 2014, with a more
advanced version that will support joint
encoding of texture and depth information coming in the early part of 2015.
Scalability is a key attribute of any
codec, as it enables trimming of video
streams to suit different network conditions and receiver capabilities; scalable extensions to HEVC are planned
for July 2014. Range extensions, which
support several color formats as well as
increased bit depths, are another area
currently under development.
In addition to these extensions, further improvements are expected to take
place inside the current HEVC framework, such as more efficient encoding
and decoding (both software and hardware). It is likely that the full potential
of HEVC will take some time to unfold,
as encoding algorithms develop and the
37
challenge posed by the optimization of
encoders and decoders in multi-core
architectures is overcome.
In short, HEVC or H.265 is twice as
efficient as its 10-year-old predecessor,
H.264. The improved efficiency that this
codec brings will help to ease traffic load
in networks and enable the creation of
new and advanced video-based services.
The codec supports parallel processing and even though it is more complex
from a decoding perspective, tests have
shown that it is suitable for adoption in
mobile services. Compression of mobile
video streams and OTT content are the
most likely initial candidates for application of the codec, and within a few years
it will undoubtedly bring UHDTV into
our homes.
Per Fröjdh
Andrey Norkin
is director of media
standardization at
Ericsson and former head
of visual technology at
Ericsson Research. He holds an M
­ .Sc.
in engineering physics and a Ph.D. in
theoretical physics from Chalmers
University in Gothenburg, Sweden. Part
of his Ph.D. work was carried out on
scholarship at Imperial College London,
UK. Following postdoctoral
appointments in the US, and Denmark,
he held the position of professor of
theoretical physics at Stockholm
University, Sweden. He joined Ericsson
in 2000 as manager of video research
and standardization. He has
contributed to MPEG and ITU work on
H.264 and HEVC, served on the
advisory committee for the W3C, and
has been the editor of 15 standards on
streaming, file formats, and multimedia
telephony in MPEG, ITU, 3GPP and
IETF.
is a senior researcher at
Ericsson Research, Kista,
Sweden. He holds an
M.Sc. in computer
engineering from Ural State Technical
University, Yekaterinburg, Russia and a
Ph.D. in signal processing from the
Tampere University of Technology, in
Finland. He has worked at Ericsson
Research since 2008, contributing to
HEVC standardization through
technical proposals and activities,
including the coordination of a core
experiment on deblocking filtering,
chairing break-out groups and
subjective quality tests for the Joint
Collaborative Team on Video Coding
(JCT-VC). He has also been active in the
3D video standardization for JCT-3V.
He is currently the project manager of
the 3D VISION project at Ericsson
Research, working on 3D video
systems, and algorithms, as well as on
parts of the standardization.
Rickard Sjöberg
References
1. ITU, January 2013, press release, New video
codec to ease pressure on global networks,
available at: http://www.itu.int/net/pressoffice/
press_releases/2013/01.aspx#.UWKhxBnLfGc
2.Ericsson, November 2012, Mobility Report,
available at:
http://www.ericsson.com/
ericsson-mobility-report
3.Fierce Broadband Wireless, 2013, Ericsson CEO:
90% of network traffic will be video, available
at: http://www.fiercebroadbandwireless.com/
story/ericsson-ceo-90-network-traffic-will-bevideo/2013-02-25
4.ITU-T Recommendation H.265 | ISO/IEC 230082: High Efficiency Video Coding, available at:
http://www.itu.int/ITU-T/recommendations/
rec.aspx?rec=11885
5.IEEE, December 2012, Overview of the High
Efficiency Video Coding (HEVC) Standard,
available at: http://ieeexplore.ieee.org/stamp/
stamp.jsp?tp=&arnumber=6316136
6.IEEE, December 2012, HEVC Deblocking Filter,
available at: http://ieeexplore.ieee.org/stamp/
stamp.jsp?tp=&arnumber=6324414
7.IEEE, December 2012, Overview of HEVC
High-Level Syntax and Reference Picture
Management, available at: http://ieeexplore.ieee.
org/stamp/stamp.jsp?tp=&arnumber=6324417
8.IEEE, DOCOMO Innovations, December 2012,
HEVC Complexity and Implementation Analysis,
available at: http://ieeexplore.ieee.org/xpl/
articleDetails.jsp?arnumber=6317152
9.Ericsson, 2012, Ericsson announces world’s
first HEVC encoder for live TV delivery to mobile
devices, available at: http://www.ericsson.com/
news/120822_ericsson_announces_worlds_
first_hevc_encoder_for_live_tv_delivery_to_
mobile_devices_244159018_c
is a senior specialist in
video coding in the
Multimedia Technologies
department at Ericsson
Research, Kista, Sweden. With an M.Sc.
in computer science from the KTH
Royal Institute of Technology,
Stockholm, Sweden, he has been
working with Ericsson since 1997 and
has worked in various areas related to
video coding, in both research and
product development. In parallel, he
has been an active contributor in the
video-coding standardization
community, with more than 100
proposals relating to the H.264 and
HEVC video-coding standards. He is
currently working as the technical
leader of Ericsson’s 2D video-coding
research, including HEVC and its
scalable extensions. His research
interests include video compression
and real-time multimedia processing
and coding.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
The merger of two giants
38
Next generation
OSS/BSS architecture
Breaking down the silos of operations and business support systems (OSS/BSS) to form an
integrated, cross-functional platform that can take a product from conception to execution in
a simplified and consistent manner will cut time to market from months and years to weeks.
JA N F R I M A N, L A R S A NGE L I N, E DVA R D DR A K E A N D M U N I S H AGA RWA L
The systems that keep networks
running and profitable are in the
direct line of fire when it comes
to implementing change. So, as
the world moves toward global
connectivity, as smartphones
cause a shift in user behavior,
and as subscribers demand more
personalized products and even
greater control, the functions
of OSS/BSS – such as planning,
configuration, fulfillment,
charging, billing and analytics –
need to be integrated.
A consolidated architecture is a typical
computer-science approach for bringing
together the functions of different systems. By adopting such a consolidated
architecture for OSS/BSS, operators will
be able to maintain control over costs
while implementing network changes
effectively.
BOX A The challenges of evolution
By exposing the functionality and information held in their networks, operators have the opportunity to create
innovative and ever more complex value chains that include developers, OTT
players and subscribers. In these new
value chains, the flow of information
and control shifts from unidirectional to multidirectional, and participants
can be consumers of services and information as well as being producers of
them.
New business models for network evolution are based on providing anything
as a service (XaaS) – including IaaS, PaaS,
SaaS and NaaS – and when using this
model, it is not just value chains that
become more complex; the life cycles
of products and services also become
more diversified.
How then, as business models
advance, should OSS/BSS requirements
evolve to cater for factors such as big
data, personalization and virtualization? The simple answer is through
configurability. To create a high level
of flexibility, the evolution of OSS/BSS
needs to be configuration driven, with
an architecture based on components.
The impact of big data
Information is a critical resource. Good
information is a key asset – one that can
be traded, and one that is critical for
optimizing operations. As volumes rise,
the rate of creation increases, and a wider variety of data that is both structured
and unstructured floods into OSS/BSS,
access to storage needs to be effortless.
In this way, tasks and optimization processes can maximize the use of existing
infrastructure and keep data duplication to a minimum.
Data management needs to be secure
and controllable, ensuring that the
Terms and abbreviations
BO business object
BPMN
Business Process Model
and Notation
BSS
business support systems
CEP complex event processing
CLI
command-line interface
(E)SP
(enterprise) service bus
ETL
extract, transform, load
eTOM
enhanced Telecom Operations Map
GUI
graphical user interface
IA
information architecture
IaaS infrastructure as a service
IM
information model
JEE
Java Enterprise Edition
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
LC
life cycle
LDAP
Lightweight Directory
Access Protocol
M2Mmachine-to-machine
NaaS
network as a service
NFV
network functions virtualization
OLAP
online analytical processing
OLTP
online transaction processing
OS
operating system
OSGi
OSGi Alliance (formerly Open Services Gateway Initiative)
OSS
operations support systems
OTTover-the-top
PaaS
platform as a service
PO
purchase order
RAM SaaS SBVR
SDN
SID
SLA
SQL
TCO
TMF UI
VM
XaaS
random access memory
software as a service
Semantics of Business Vocabulary and Business Rules
software-defined networking
shared information/data model
Service Level Agreement
Structured Query Language
total cost of ownership
TeleManagement Forum
user interface
virtual machine
anything as a service
39
systems accessing information do not
jeopardize data integrity and subscribers can feel confident that their information is protected.
FIGURE 1 Business
application
plane
Business application
(such as network inventory)
Control
plane
SDN
controller
Data
plane
ND
Business application
(such as VPN service)
SDN
controller
ND
ND
ND
ND
SDN
controller
ND
ND
ND
ND
SDN
controller
ND
ND
ND
ND
ND
ND
SDN architecture
The impact of M2M
As the number of connected devices
gets closer to 50 billion, the need for
automated and autonomous behavior
in processes such as configuration and
provisioning is becoming more significant. Being able to remotely configure,
provision and update millions of devices without impacting the network supports scaling while maintaining control
over opex.
Making good use of technology
One way to address these challenges is
to make good use of advancing technology, particularly when it comes to
OSS/BSS implementation architecture.
And it’s not just about using technology development in a smart way; it’s
also about understanding the potential
of a given technology. So, when a new
concept results in a significant breakthrough, the services and products that
Business application
(such as CRM)
Business application
(such as charging and billing)
The impact of subscriber needs
Personalized services and superior
user experience are key capabilities
for business success and building loyalty. Subscribers want to be in control,
and feel that their operator provides
them with reasonably priced services
that meet their individual needs, over
a network that delivers near real-time
response times. The ability to create and
test services in a flexible way with short
time to market will help operators meet
changing user demands.
The impact of virtualization
As a result of virtualization, operators,
partners and even subscribers (in the
future) can create instances of their services and networks on demand. So, as
networks continue to move into the
cloud, and SDN and NFV technologies
become more widespread, the number
of entities managed by OSS/BSS will rise
by several orders of magnitude.
So, to help operators remain competitive, next generation OSS/BSS need to
fully address the challenges created by
certain aspects of network evolution,
including virtualization, big data, M2M
and personalization.
Separating planes in SDN architecture
ND = network device
can be created as a result should be readily definable.
Capitalizing on increased flexibility
and agility made possible by new technologies (such as virtualization and
SDN) needs to be coordinated through
a management function, which puts
new demands on OSS/BSS architecture.
FIGURE 2 The evolution of virtualization
The demands created by increasing virtualization of data centers, not just in
terms of computational capacity, but
also in terms of storage and networking
capabilities, are:
virtualization of the OSS/BSS, and
running these systems in the cloud;
Abstraction of a typical OSS/BSS deployment
Enterprise’s business
Gap/disconnect
Present BSS/OSS
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
Traditional
BSS/OSS
function
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
The merger of two giants
40
FIGURE 3 Extracting hard-coded business logic
Today
Tomorrow
Enterprise’s business
Enterprise’s business
Gap/disconnect
Enterprise’s virtualized BSS/OSS
Present BSS/OSS
Traditional BSS/OSS
app
Business rules
Business processes
Business events
Information
.......
Integration
Storage
Analytics
BSS/OSS
function
BSS/OSS
function
management of cloud-based OSS
applications such as service assurance;
and
management of cloud-based BSS
applications such as IaaS and PaaS.
It may, however, not always be beneficial to run certain network elements
on generic IaaS resources. For example,
information stored in a database may be
better provided in the form of a service
to subscribers in an IaaS environment,
rather than as virtually deployed tenants. The general rule is that anything
provided as a service, which is implemented by a piece of software running
in a generic IaaS environment, has a
reduced level of control and ­efficiency.
Due to the extra layers created by running software in a generic environment, the drawbacks of this approach
must be weighed carefully against the
benefits of increased flexibility and better (shared) use of physical resources.
For next generation OSS/BSS, the
focus should be placed on implementing
flexibility in an efficient way ­together
with automation and orchestration of
resource allocation.
The hypervisor approach to
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Business
object
Enterprise
information
model
Ericsson BSS/OSS
application logic
Ericsson BSS/OSS
application logic
Traditional
BSS/OSS
function
Ericsson BSS/OSS
application logic
Ericsson BSS/OSS
application logic
Traditional
BSS/OSS
function
Business
rule
Shared
information
model
Business
process
Business
event
Common
storage
Common
analytics
virtualization, where virtual machines
(VMs) share the resources of a single
hardware host, is evolving so network
infrastructure is becoming more efficient. For example, the failover capabilities of the hypervisor can place agents
on the host in a similar way to traditional failover clusters, and can monitor not only VM health, but application
and OS health as well. Such features
are prerequisites of an efficient ­virtual
environment. However, application
architecture may have to take these
features into account, as in some cases
they cause the responsibility to perform
certain tasks (such as data recovery) to
shift between the application and the
infrastructure.
Service provider SDN
SDN separates the data plane (the forwarding plane or infrastructure layer)
from the control plane, which in turn,
is separated from the business application plane. As shown in Figure 1, various business applications communicate
with SDN controllers, providing a virtualized – possibly hierarchical – view of
the underlying data plane.
Generally speaking, the management
Common
integration
requirements for SDN and non-SDN
architectures are similar, if not the
same. For example, both require inventory, ordering and fault management.
However, SDN presents a new set of technical issues related to resource management, which brings into question the
current partitioning and structure of
OSS/BSS architectures.
Specifically, SDN can result in horizontally abstracted virtualized software
layers that have limited vertical vision
through the hierarchy from the business applications to network devices.
So, at the same time as the abstraction
offered by SDN makes it easier to expose
the capabilities of the network, it creates
additional challenges for the OSS/BSS
architecture.
OSS always needs to have the capability to map the virtual view of the
network to the underlying implementation. Sometimes, the SDN controller
handles domain-specific OSS/BSS functionality by hiding parts of the complexity of the control and data planes;
and as a result, only a subset of information will be propagated to the business application plane. Sometimes,
the underlying layers are not even visible – such as when a third party owns
them. In these cases, SLAs can be used
to map the virtual view to the underlying implementation.
The evolution to SDN architecture
and virtualization causes the number
of entities managed by OSS/BSS components to rise, which in turn impacts the
way they are managed. For example, a
more extensive history of each e
­ ntity is
required, because the semi-­static environment used to locate a device using
its IP address no longer exists. With
SDN, network topology becomes totally dynamic. History data is essential for
analysis and management of the network, as this information puts network
events into context.
Hybrid flexibility
Modern database design is evolving
toward the use of hybrid architectures.
This approach allows a wider range of
solutions and applications to be created
with a single consistent implementation and one logical data store.
Hybrid disk/in-memory databases
use in-memory technologies to achieve
the performance and low latency levels
41
of an in-memory solution, while still
using disk for data persistency. The
hybrid approach allows more data to
be stored on disk than can fit into memory; as such, the disk is not a mirror of
the in-memory content. This approach
is similar to caching disk content, while
providing the performance that comes
from a true in-memory design – which
cannot be achieved by caching disk content alone.
Hybrid SQL/NoSQL (sometimes
referred to as NewSQL) solutions are
SQL-capable databases that are built
using a NoSQL implementation to attain
the scalability and distribution that
these architectures afford, while still
providing support for SQL. However,
such hybrid solutions can be limited by
their lack of support for ­partition-wise
joins and subsequent lack of support
for ad hoc queries – although there are
exceptions to this. Hybrid OLTP/OLAP
solutions aim to merge the typical
characteristics of transactional OLTP
workloads and OLAP-based workloads
(related to analytics) into a single implementation. To build such a structure
typically requires that both database
architectures be considered from the
outset. Even if such solutions exist, it
is difficult to build this type of hybrid
from the starting point of an OLTP- or
OLAP-optimized architecture.
Big data
Modern data centers are designed so
that increasingly large amounts of memory with low levels of latency are being
placed ever closer to ­computational
resources. This greatly increases the
level of real-time processing that can be
achieved as well as the volumes of data
that can be processed. Achieving these
processing levels is not simply a matter
of the speed at which operations can be
carried out; it is also about creating new
capabilities. The developments being
made in big-data processing have a significant impact on how next generation
OSS/BSS architecture can be designed.
Fast data – the velocity attribute of big
data – is the ability to make real-time
decisions from large amounts of data
(stored or not) with low latency and fast
processing capabilities. Fast data supports the creation of filtering and correlation policies that are based on – and can
also be adjusted to – near real-time input.
Another big-data concept combines the in-memory/disk hybrid
with the OLTP/OLAP (row/column)
hybrid to achieve a single solution
that can address both OLTP and
demanding analytics workloads.
Coupled with the huge amounts of
memory that modern servers can
provide, this approach removes
the need for a separate analytics
database.
The business logic
When OSS/BSS are deployed, they
bring business and technical stakeholders together and allow them to
focus on the design and implementation of their unique business. The
functionality provided by OSS/BSS
must support the necessary userfriendly tools to implement and
develop business logic. The deeper
and more flexible this support is, the
more business opportunities can be
explored, and the more profitable an
enterprise can be.
Figure 2 shows an abstraction of a
typical OSS/BSS deployment. Current
implementations tend to be multivendor, with multiple systems performing
similar tasks. Organic growth has led to
a lack of coordination and as a result, significant time and effort is spent on integration, time to market is long, and TCO
tends to be high. Quite often there a significant gap between daily business and
the systems used to support it.
To transform such a complex architecture into a more business-agile
system requires some evolution. As consolidation is fairly straightforward, this
tends to be the first step. However, to
succeed, it requires systems to be modular, to be able to share data, to use data
from other sources, and to have a specific role. Another approach is to reduce
the silo nature of OSS/BSS and instead be
more flexible using business logic in
FIGURE 4 An order handling process. The notation shown here is conceptual;
processes are modeled in BPMN1 2.0 and rules in SBVR 2
Receive
Send
purchase order purchase order
reception
acknowledgement
External
Purchase
order
Required
entities
Order
handling
Customer
handling
Send
purchase order
delivered
Product
offering
Activate customer order
Event and Purchase
process
order
received
Created BO
and LC
Application
tasks
Create
customer
order
Check
customer
order
Customer
order
Customer
order
Activate
services
Activate
resources
Activate
billing
Archive
customer
order
Customer
order
Activate
service
Activate
resources
Activate
billing
Archive
customer
order
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
The merger of two giants
42
FIGURE 5 The information model is the fundamental component of Ericsson’s
approach to next generation OSS/BSS.
Business logic from conception to execution
Strategy
Design
Implement
Enterprise’s business studio
Deploy
Enterprise’s
BSS/OSS
management
C-level
management
Business
architects
Enterprise
and IT
architects
Business
logic
decision
Business
logic
design
Business
logic
implement
Operate
Enterprise’s
BSS/OSS
engines
Business
logic
management
Executable
business
logic
Architecture proposal
As visualized in Figure 3, the Ericsson
approach to next generation OSS/BSS
is to extract hard-coded business logic
FIGURE 6 from the underlying systems, and to
structure functionality according to
design and life-cycle flow. To achieve
this and build an abstract and ­virtual
view of the business successfully, a
common, shared and semantically rich
information model (IM) and a defined
set of relationships are essential.
Information models within an enterprise
Enterprise OSS/BSS architecture reference model
Other
enterprise systems
Deployed OSS/BSS architecture
Partner
Enterprise
business concepts model
Enterprise’s application
level information model
Other
Other
Other
application level
information
model
Enterprise data model
Ericsson application
level information model
Mediation/
transformation
Enterprise production
environments
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
actors and roles – such as companies,
functions, individuals and customers,
suppliers and service providers;
services and functions – such as sales
and contracting;
processes – such as TMF eTOM;
business objects – such as products,
orders, contracts and accounts; and
rules – such as pricing, prioritization and
product/service termination.
These building blocks can then be used
to form processes, as Figure 4 illustrates. To achieve the full degree of
flexibility, the building blocks for the
complete life cycle (from definition to
termination) are needed.
Feedback
a configurable manner. Either way,
the best evolution approach is in system
architecture design.
Building blocks
When the hard-coded business logic is
extracted the following building blocks
are created:
Transformation
Conception to execution
Figure 5 illustrates design chain functionality. This process is relied upon to
take a new business idea from conception to execution.
Business logic is defined, designed
and implemented in what Ericsson
refers to as the enterprise business studio. The studio comprises a set of integrated workbenches that have access
to all building blocks and information elements, and provide feedback to
the process owner as business logic is
implemented.
In the Ericsson model, business logic (such as verification, commissioning
and decommissioning, supervision,
migration and optimization) is transferred to a management function for
implementation and application. The
management function is also responsible for transferring business logic to the
proper execution engines. For example,
pricing rules are used in execution by
the rating engine, contracting engine
and the sales portal engine. Given the
potentially massive spread of pricing
rules, getting it right at this stage of
development is key.
Information architecture
Generally speaking, an enterprise
defines the information it needs to
43
operate, and the OSS/BSS manage this
information, based on a range of business models. The spread of information
across any given enterprise is extensive,
and can span many different functional
areas from marketing, ordering, strategy and HR, to production and finance.
Information models used in OSS/BSS
include:
As operators continue to differentiate
and offer ever more complex products
and services, the requirements on information change. Information is no longer
just mission critical; it is also enterprise
critical, and changes constantly as business needs evolve.
The shift to next-generation OSS/
BSS changes the way enterprise-­critical
information needs to be handled, creating a number of system requirements:
information and applications need to be
separated;
the entire life cycle – from definition to
termination – needs to be modeled;
information needs to be shared among
all enterprise users;
master data needs to be determined,
and even multiple masters need to be
supported to align with different
enterprise functions; and
information needs to be characterized in
terms of size, throughput, quality,
reliability and redundancy across the
board, for all functions and applications
– one instance that can be used by all.
As Figure 6 shows, information held
in an enterprise can be generated by
many sources – both internal and external – and can be classified according to
its properties and type. For example,
information can be static, structured,
event-driven or transactional.
The best way to meet the new
Information architecture implementation
Enterprise
Enterprise
OSS/BSS
information
management
Data access and
grid services
Master data
Reference and ID
Application Customer
and partner
layer
management
Non-real-time API
Enterprise
catalog
Event
Analytics
management management
Data access and grid services
Real-time API
Data grid framework
Communication buses
Data access
IM
Metadata
functionality
Data transform
Information
model
Transaction
support
Information access
(E)SB
distribution
Published
IM
Reference
and global ID
Information service
registry
Integration
master/slave
External data sources
connection
Data engine
services
Analytics and SEP
engines
Non-real-time API
Data engine
framework
Data persistance
Grid and vault
Data engine services
Non
Ericsson
apps
Bi-directional integration
and transformation
services
the enterprise vocabulary and concepts
at the business level – for example, an
enterprise might refer to a voice product
using its marketing name, such as
Family and Friends;
the canonical concepts at the
application level – which might refer
to the Family and Friends product as
family-group; and
the multivendor concepts at the
application level – where the concept
of Family and Friends has different
names at the business level and the
canonical level.
FIGURE 7 Real-time API
Analytics and CEP
engines
Storage schemas
engine
Data vault
requirements on information – ­driven
by the need to differentiate – is to design
OSS/BSS in a way that is independent of
functional applications with a centrally
managed information architecture (IA)
that has a common and shared information model.
The key characteristics of this architecture are:
integrated information across functions;
information offered as a service –
facilitating high-level abstraction and
avoiding the need to understand lowlevel data constructions; and
information published in catalogs –
formalizing IaaS and enabling use by
process owners.
Due to the complexity and widespread
nature of information models, a modularized information architecture is
needed – one that can be configured to
meet the varying needs of enterprises
and used in multivendor scenarios with
varying information life cycles.
Modular layers
A modular IA can be implemented by
categorizing information into a matrix.
Storage
The first step is to categorize information into (horizontal) layers, where each
layer is populated by a number of entities. Subsequently, these entities can be
combined (vertically) into a solution.
As Figure 7 shows, the data vault is
placed at the bottom of the architecture hierarchy. The most efficient type
of storage can be chosen from a number
of components, including disk, RAM or
virtual resources.
A level up from the data layer is the
engine layer, which comprises a set of
components that provide access to the
information storage. SQL, LDAP, NoSQL
or HBase are examples of technologies
used in this layer, and each group of data
selects the technology that best ­matches
the access requirements for that data.
The grid layer, above the engine, is
where the information model becomes
accessible. Here, the IM is divided into a
set of responsibility areas, such as enterprise catalog with product, service and
resource specifications or inventory
data. These areas deploy components
that expose information as a service
and protect the underlying data,
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
The merger of two giants
44
ensuring that it is consistent and
available for all authorized applications.
The applications that consume and
produce information sit on the top of
the architecture. They set the requirements on the grid layer so that information is made available with the right
characteristics and accessibility according to the needs of the given application. Applications can of course use local
caching, but the data vault allows information to persist and be made available
throughout the system.
The right-hand side of Figure 7 shows
the set of functions that support the flow
of information in and out of the model.
Typically, this part of the OSS/BSS interfaces to data owned by legacy or other systems, and allows information to
be accessed from the outside. Grid components are responsible for interacting
with external data sources, and exposing access to information residing in the
model. Typical functions include transformation, protocol adaptations, handling of services or data streams and
identity mapping, which are also used
by applications in the application layer.
FIGURE 8 Management functions such as definition, registration, discovery, usage,
archiving and decommissioning are
shown on the left of the information
architecture. As information is no longer hard-coded in each application,
but shared among applications at all
stages of the life cycle, the management function is vital for ensuring data
consistency.
By making use of the common services provided by a deployment stack
that supports scale-out architecture and
meets the needs of big data, application
development should become more efficient. The stack should integrate easily with existing enterprise systems – a
capability that becomes more significant as OSS/BSS are developed and used
in multiple scenarios around the world
and deployed in an IaaS manner.
Deployment stack
To serve next generation OSS/BSS, a
state-of-the-art deployment stack is
required. A functional view of such
a stack is illustrated in Figure 8. The
deployment stack should provide a consistent user experience for all processes
– from business configuration and system provisioning, to operations, administration and management. It should
support applications deployed on a variety of different infrastructures including cloud, and virtualized and bare
metal hardware. The deployment stack
should provide a means of efficient integration among applications, and enable
service exposure in a uniform way.
Hardware
Using existing hardware infrastructure for OSS/BSS deployment is the best
option for operators as it consolidates
the use of hardware, and supports rapid
reconfigurability and scalability. Linux
is an attractive OS, as it is community
driven and supports the decoupling of
software and hardware elements.
A functional view of the deployment stack for next generation OSS/BSS
Presentation
Graphical user interface
Command line interface
Machine-to-machine interface
Business level
Applications
Information tier
Application services
Operations and
management
Middleware
In-service performance
Logging
JEE/OSGi
IP
OS
Licensing
Configuration management
Availability support
User management
Coordination service
Performance management
Software management
Fault management
Backup and restore
Load balancer
Linux
Virtualization
Hardware
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
File system
Cloud
deployment
Middleware
A common and well-composed middleware provides: a consistent environment, effective management, ease of
integration, greater availability, better scaling, load balancing, simplified
installation, upgrade and deployment,
and improved backup capabilities.
Such an environment can be provided by either an OSGi container or a
JEE application server pre-integrated
with availability management, software management and backup/restore
capabilities.
Operations and management
This function provides common management services for configuration, logging, configuration management, and
fault and performance management.
Application services
This layer provides common functions
for OSS/BSS applications, such as:
service performance – which monitors
and reports uptime;
licensing – which enables provisioning,
monitoring, control and reporting of
licenses;
user management – which provides
authentication and authorization; and
coordination service – which provides
inter-application coordination in a
distributed environment that supports
changing license requirements created
by business models such as
pay-as-you-grow.
45
OSS/BSS
Presentation layer Common operations
and management GUI
Software management
GUI/CLI
Business studio
UI
Business level
Application level
Debt
management
Revenue
management
Experience and
assurance
CEP
Event
handling
Decision
support
Communication
channels
Correlation
Analytics
Resource
management
Order
management
Workflow
framework
Document
format
Content
management
CEP and ETL
management
Common
application
functions
Rule
framework
Application
SW compliance
and
governance
Enterprise
catalog
App IM and
master data
governance
Customer and
partner
interaction
App service
governance
BSS/OSS
application
functions
Customer and
partner
management
App level
management
Business studio
Proposed architecture
The architecture of next generation
OSS/BSS is illustrated in Figure 9. At
the business level, the proposal supports service agility in all processes
from conception to retirement and all
relevant phases – including planning,
deployment, customer on-boarding and
assurance.
The proposed architecture comprises a set of application functions, which
implement tasks such as enterprise
catalog, charging, billing, order management, experience and assurance.
Application functions are implemented through a set of components that can
be configured and assembled so they
form a complete solution that can also
be integrated with existing systems.
The set of common application functions, including correlation and event
handling, support the specific OSS/BSS
application functions.
The information architecture separates the information model so information is matched to the application
functions, supporting modularity and
enabling integration in the overall information model. The information model
is based on a shared information/data
model (SID), with extensions embracing more standard industry information models.
To support effective implementation,
all components should be ­prepared for
cloud deployment.
Next generation OSS/BSS architecture
FIGURE 9 Application information mode
Presentation layer
The presentation layer provides support
for GUI, CLI and M2M interfaces for OSS/
BSS applications. A common GUI framework, together with single sign-on for
the entire stack, is key to providing a
consistent user experience.
By exposing interfaces to other applications in a uniform way, the amount
of application-application integration
required is reduced significantly.
Application level information tier
Application services
Operations and management
Middleware
IP
OS
Virtualization
Hardware
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
The merger of two giants
46
References
1. Business Process Modeling
Notation, available at: http://www.
bpmn.org
2. Semantics of Business Vocabulary
and Rules, available at: http://
www.omg.org/spec/SBVR
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Jan Friman
Munish Agarwal
is an expert in the area of
user and service
management at Business
Unit Support Systems
(BUSS). He has held
various positions in the area of OSS/
BSS at Ericsson for 16 years, including
R&D, system management and
strategic product management. He is
chief architect for information
architecture at BUSS and holds an
M.Sc. in computer science from
Linköping Institute of Technology,
Sweden.
is a senior specialist in
multimedia architecture
and chief implementation
architect for OSS/BSS. He
has been at Ericsson since 2004,
working in the OSS/BSS area. He is
currently driving the BOAT
implementation architecture and is the
product owner for Next Generation
Execution Environment. He holds a
B.Tech. in material science from the
Indian Institute of Technology,
Kharagpur, India.
Edvard Drake
Lars Angelin
is an expert in the area of
hardware and software
platform technologies,
and is chief architect for
implementation architecture at BUSS.
He has 20 years’ experience at
Ericsson, ranging from AXE-10
exchanges to today’s commercial and
open source technology innovation. He
holds a B.Sc. in software engineering
from Umeå University, Sweden.
is an expert in the
technology area
multimedia management
at BUSS. Lars has more
than 28 years of work experience in the
areas of concept development,
architecture and strategies within the
telco and education industries. Lars
joined Ericsson in 1996 as a research
engineer, and in 2003 he moved to a
position as concept developer for telconear applications, initiating and driving
activities, most of them related to M2M
or the OSS/BSS area. He holds an
M.Sc. in engineering physics and a
Tech. Licentiate in tele-traffic theory
from Lund Institute of Technology,
Sweden.
Re:view
47
Nine decades
of innovation
Automatic exchanges to smart networks
One of the most significant technical evolutions in
the history of telecoms is that of the mobile phone. In 1990,
Ericsson received its first order to supply a GSM network,
which brought the company’s switching and radio expertise
together. At the same time, the internet began its worldwide
expansion, as did the liberalization of the telecom market.
In 1993, Sweden became the first European country to
deregulate its telecom market. While already a fact in the US,
deregulation has led to increased competition and a greater
need for innovation to deliver customer benefits.
In 1994, after almost a decade of research, Ericsson released
the Bluetooth technology standard, which allows devices to
exchange information wirelessly. In the spirit of open standards, its control has been handed over to the Bluetooth
Special Interest Group (SIG) – a non-profit organization.
That same year, Ericsson Review carried an article about
intelligent network architecture for the Japanese 2G digital cellular standard Personal Digital Cellular (PDC). This
enhanced network architecture was based on the principle of
strictly separating network-oriented services from the mobile
subscriber-specific services, allowing for the rapid development of new services in response to demand.
At the turn of the 21st century, Ericsson addressed the challenges of VoIP. The surpassing of voice by data – a fact that
didn’t become reality until the end of 2009 – was clear, and
the primary challenge was how to port voice services to the
new packet-based platform, while maintaining the same level of quality.
The first decade of the new century was dominated by LTE
and the desire to evolve network architecture to support technology evolution, improve spectral efficiency, be more flexible
and ultimately support new services and superior user experience. In February 2007, Ericsson demonstrated LTE with bit
rates of up to 144Mbps, and theoretical peak rates of 1.2Gbps
were demonstrated in 2010. In 2009, Ericsson delivered the
first commercial LTE network.
And last but not least, the dramatic impact on networks
created by smartphones, tablets and other mobile devices,
the increased demand for mobile broadband and soaring
data traffic have ramped up the opportunities in telecoms.
Ericsson today holds a unique position in that we can provide
network equipment, solutions, services, and management
tools that encompass the full spectrum of needs.
Looking forward, one of our visions for the Networked
Society is that everything that can benefit from a connection
will be connected. Connectivity is the starting point for new
ways of innovating, collaborating and socializing.
Ultra modernism
In 1984, an entire issue of
Ericsson Review was
dedicated to fiber, providing a
report on fiber optic activities
at Ericsson. The articles in this
issue covered everything from
cable design, optical-fiber
transmission, splicing
equipment and the use of
semiconductor materials that
were more advanced than
silicon. One of the articles
Ericsson Review,
featured the application of
F issue, 1984.
fiber optics in offshore
systems. This type of
environment posed tougher requirements on fiber design
in terms of safety, weight and transmission distances.
The age of the internet
In 1998, an entire issue of
Ericsson Review was
dedicated to the internet and
its impact on different aspects
of the telecoms industry.
Scalability was the name of
the game as were solutions
that optimized the use of
network resources. The focus
of the articles in this issue was
access, as at the time, most of
the costs were hidden in this
part of the network. In the
wake of liberalization, the
Ericsson Review,
access network was deemed
special internet issue,
to be the battleground for
1998.
competing offerings in terms
of bandwidth and service levels.
A new millennium
In 2007, Ericsson Review
carried an article about LTESAE (long term evolutionsystem architecture
evolution). The simplified and
optimized architecture uses a
minimum number of nodes in
the user plane. In addition,
new features have been
introduced to simplify
operation and maintenance.
Ericsson Review,
issue 3, 2007.
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Switching the smart way
48
Carrier Wi-Fi:
the next generation
By controlling whether or not a device should switch to and from Wi-Fi, and when
it should switch, cellular operator networks will be able to provide a harmonized mobile
broadband experience and optimize resource utilization in heterogeneous networks.
RU T H GU E R R A
Next generation carrier Wi-Fi will
overcome existing coordination
issues in multi-RAT environments
to become an integrated
component of mobile broadband
offerings. Guaranteeing the best
mobile broadband experience
and ensuring that resources in
a heterogeneous network that
includes Wi-Fi are utilized in an
optimal way, is only possible
if subscribers are connected
to Wi-Fi when this is the best
option for them and for the entire
network. While this may sound
obvious, the way subscribers
currently switch to and from Wi-Fi
is not optimal. Today, the decision
to connect to Wi-Fi is taken by
the device according to one basic
principle: if Wi-Fi is available, then
use it for data traffic.
However, this approach is short-sighted because it does not take into consideration real-time information about all
BOX A AES
AKA
ANDSF
AP
BSC
BSS
CSMA
DAS
DPI
EAP
GTP
the available resources. In a heterogeneous network, resources can include
2G, 3G, LTE, macro, small cells, different
carriers, different protocols (802.11g,
802.11c) as well as different channels.
In addition, devices do not take into
account the activity of other UEs, and
so each decision to switch to or from a
Wi-Fi network is made independently,
and without any consideration for load
balancing. In short, current practice is
inherently inefficient.
Background
The smartphone revolution and the
near-ubiquitous support for Wi-Fi of
modern devices have created new business opportunities and new challenges
for telcos. Operators have so far deployed
over 2 million Access Points (APs) in public spaces, and there are currently about
8 million hotspots worldwide. But why
is there a need to integrate Wi-Fi into
operator networks? The simple answer
is that this technology is a good complement to existing solutions, and in
certain conditions, it is particularly
appropriate for handling spikes in data
traffic. But to work well, it needs to be
integrated.
So, the main factors that next generation Wi-Fi will be able to capitalize on
are:
the vast amount of unlicensed spectrum
that can be used by this technology
without the need for any regulatory
approval;
its ability to offload data traffic,
complementing existing indoor
solutions, such as small cell and DAS;
the near-ubiquitous device support for
this technology – including UEs that are
non-cellular;
the evolution of small cells to support
both cellular and Wi-Fi technology; and
a new level of maturity – exemplified by
the development of, and device support
for, new standards and products, such
as Hotspot 2.0, 802.11ac, EAP
authentication and additional solutions
currently being defined, such as S2a
mobility, and 3GPP/Wi-Fi integration.
Terms and abbreviations
Advanced Encryption Standard
Authentication and Key Agreement
Access Network Discovery and Selection Function
Access Point
base station controller
basic service set
carrier sense multiple access
Distributed Antenna System
deep packet inspection
Extensible Authentication Protocol
GPRS Tunneling Protocol
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
HLR
HSS
LI
MME
MS-CHAP
MWC
NAS
PLMN
RAT
RF
RNC
home location register
Home Subscriber Server
Lawful Interception
Mobility Management Entity
Microsoft’s Challenge-
Handshake Authentication Protocol
Mobile World Congress
non-access stratum
Public Land Mobile Network
radio-access technology
radio frequency
radio network controller
RRM
Radio Resource Management
RSRP
reference signal received power
RSSI
received signal strength indicator
SON
self-organizing networks
TCP
Transmission Control Protocol
TLS
Transport Layer Security
TTLS
Tunneled Transport Layer Security
UE
user equipment
USIM
Universal Subscriber Identity Module
WIC Wi-Fi controller
WLAN
wireless local area network
49
The results of a survey1 carried out
with 24 service providers are illustrated in Figure 1. These highlight the
challenges operators are focusing their
attention on when it comes to carrier
Wi-Fi deployment.
Some steps have already been taken
to include Wi-Fi in mobile broadband
solutions, such as EAP authentication.
Some solutions are already supported by
UEs, while others will be available shortly. But much more can be done.
With these challenges in mind, the
top three priorities for next generation
carrier Wi-Fi are:
traffic steering 3GPP/Wi-Fi – to
maintain optimal selection of an access
network so quality of experience can be
ensured and data throughput
maintained;
authentication – to provide radio-access
network security for both SIM- and nonSIM-based devices; and
DPI, support for unified billing and
support for seamless handover –
achieved by integrating with the core
infrastructure already deployed for
3GPP access.
When the varied set of resources in a
heterogeneous network can be combined and optimized, networks can
provide increased capacity and the performance needed to give subscribers
the desired level of user experience. So
for Wi-Fi, the objective is not to turn it
into a 3GPP technology, but rather to
figure out how to add 3GPP intelligence
and control over Wi-Fi usage, so that all
resources are used in an optimal way
while delivering the best user experience. When Wi-Fi becomes just a
­ nother
RAT, the synergies and application of
mobile network capabilities, intelligence and infrastructure will remove
the burden from Wi-Fi to meet all of the
challenges outlined in Figure 1.
With the operator in control, and
with Wi-Fi networks that are integrated with mobile radio-access and core
networks, subscribers will experience
high-­performing mobile broadband
that operates in a harmonized way.
Operators will be able to control, predict and monitor the choice of connectivity, allowing them to optimize both
the user experience and resource utilization across the entire network.
FIGURE 1 Mobile and Wi-Fi network integration: the main challenges
n= 24
Challenges
Supporting seamless handover
between mobile and Wi-Fi networks
83%
Mobile data offload traffic steering
between mobile and Wi-Fi networks
79%
Supporting QoS
67%
Maintaining data throughput
67%
63%
Radio access security
Supporting seamless handover
between Wi-Fi access points
58%
Supporting SIM and non-SIM devices
50%
Integrating carrier Wi-Fi with
3G/4G small cells
50%
Authentication security
46%
Deep packet inspection and policy control
46%
Supporting unified billing across
mobile and Wi-Fi networks
42%
Generating revenue from Wi-Fi investment
38%
0
20
Traffic steering 3GPP/Wi-Fi
In a heterogeneous network, the type
and amount of resources that are available to provide mobile broadband are
quite diverse, as networks are built
using:
multiple technologies including GSM,
WCDMA, LTE and Wi-Fi;
several types of cells including macro,
small cells and APs; and
varying network capabilities including
carrier aggregation; different carrier
bandwidths; 802.11n; and 802.11ac.
To provide the best user experience
across all available resources and optimize resource utilization, the decision
of whether or not to switch to Wi-Fi or
back to cellular, and when to switch,
needs to be made according to a more
complex set of principles than simply:
if there is an available Wi-Fi with adequate signal strength, then switch to it.
The decision should take into account
all available technologies and carriers,
all visible cells, network and UE capabilities, radio conditions for each specific UE (which requires decisions to be
made in real time, as these conditions
40
60
80
100
fluctuate rapidly), and should ­ultimately
be based on a calculated performance
estimate for each UE, given the aggregation of all terminals in the area.
In a heterogeneous network it is more
desirable to move users to different technologies, carriers and layers. And so,
real-time coordination of all resources
adds an additional layer of complexity
to the Wi-Fi/cellular decision.
For example, if we assume that the
overall goal is to obtain the best UE
throughput (as this has a direct impact
on user experience), UEs that have 3GPP
carrier aggregation capabilities should
switch to Wi-Fi much later than terminals that don’t.
In another scenario involving three
UEs, the best solution is to assign the
first UE to a coordinated macro/small
cell to deliver higher throughput, to
assign another 3GPP technology or carrier to the second UE, and to allocate a
Wi-Fi access point to the third device
– as this UE has 802.11ac capabilities
and a good radio link budget in Wi-Fi.
Good coordination is required to make
such a decision, so that the best solution
is attained for all three UEs given
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Switching the smart way
50
FIGURE 2 Concept of a performance-based mobility feature
Authentication
Request to connect
Accept/reject
WIC
Accept/reject after comparison
with UE estimated throughput in LTE
UE throughput
estimate for Wi-Fi
UE throughput
estimate for Wi-Fi
RNC
AAA
MME
UE throughput
estimate for Wi-Fi
Accept/reject
after comparison
with UE estimated
throughput in 3G
Accept/reject after
comparison with
UE estimated
throughput in 2G
BSC
their individual requirements and
those of the other two devices.
The ability to assure predictable and
consistent switching behavior in heterogeneous network architectures eases network planning, makes optimal
use of available resources and ensures
good performance for all subscribers
independent of device. This implies
that a Wi-Fi/cellular decision-making
mechanism needs to be independent
of UE type, UE operating system and
UE vendor.
Crystal ball
As good predictions can prevent UEs
from switching too often from one RAT
to another, the ability to forecast network states and available capacity is a
fundamental tool for the RAN to operate effectively and to help improve the
mobile broadband experience. Avoiding
the ping-pong effect is beneficial
because the less switching among RATs,
the fewer the signals between the RAN
and the core, and the less the impact on
battery consumption in the terminal.
The RAN is the best place in the network to measure available resources
for specific UE forecasts, as this part of
the network has a complete overview of
resources, knowledge of the distributed
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
SON features, information about potential mobility decisions, as well as an
awareness of all UEs in the area.
For years, operators have deployed
mobility features both within and
between 3GPP radio access networks
that allow the network to determine
in what cell, with which carrier or by
what technology a UE should be h
­ osted.
These features make good use of all
available resources to provide everyone connected to the network with an
optimal mobile broadband experience.
Decisions are made by the network on
the basis of the available resources,
radio conditions of UEs and the current
state of the network (an aggregation of
information from the network and all
the UEs in the area).
A mobility feature
Ericsson’s concept of a mobility feature for switching to and from Wi-Fi
networks provides operators with the
capability to control when UEs connect
to Wi-Fi and when data traffic should be
switched back to 3GPP. This concept is
based on the forecast performance for
the specific UE.
By deploying a mobility feature that is
based on performance, operators will be
in a position to add Wi-Fi to their mobile
eNodeB
broadband resources, just like any
another RAT. And as the same entities
make the decisions about 3GPP mobility, 3GPP/Wi-Fi mobility and distributed
SON features, decisions are coordinated, preventing devices from looping,
and improving resource utilization as
a result of less signaling.
To ensure that the benefits of a mobility feature supporting real-time traffic
steering can reach mass market quickly, such a feature needs to be developed
to work on legacy UEs without the need
to install a client. To achieve this and to
avoid placing additional requirements
on UEs that would delay the applicability of the feature, an interface needs
to be placed between the controllers –
BSC, RNC, Wi-Fi controller (WIC) and
MME – of the different technologies.
Switching decisions need to be made
jointly by these controllers, as illustrated in Figure 2 – a simplified flow diagram for a performance-based mobility
feature.
To illustrate how the concept works,
it may be useful to consider the case of a
UE that is moving closer to an operator
AP, but is currently hosted in a 3G cellular network:
the default UE behavior is to request a
connection to the Wi-Fi on detection of
the AP;
the WIC will proxy the authentication
request;
on authentication, the WIC provides
the RNC currently hosting the UE with
an estimated performance for the UE
in Wi-Fi;
if the 3G throughput estimation is higher
than the estimates for Wi-Fi, the RNC
will order the WIC to reject the
connection, thus maintaining data
transmission over 3G.
As long as the UE is in the vicinity of
the AP, the connection manager in the
UE will keep trying to connect to the
Wi-Fi, allowing the network to decide
to switch when and if the conditions
become more favorable in Wi-Fi.
Once a client is accepted into the Wi-Fi
access network, the UE throughput in
Wi-Fi and the estimated throughput for
the UE in the NodeB are monitored continuously. When NodeB throughput surpasses Wi-Fi performance for the UE,
it is disconnected from Wi-Fi and data
communication will switch back to 3G.
51
Adding smartness
In addition to guaranteeing the best
experience for users and optimizing use
of resources, maintaining control over
the Wi-Fi/cellular switching process provides operators with a platform that can
be rapidly adapted to include additional
parameters. For example, operators will
be able to include subscription parameters or service considerations into the
switching decision, without having to
wait for support to be implemented by
all UEs.
The real-time switching decision has
a greater impact on user experience
for UEs that are in connected mode.
Consequently, an extension of existing
mobility features for connected mode
that includes support for Wi-Fi would
optimize and enhance this decision.
This could be achieved by including
Wi-Fi parameters in UE measurement
reports (which currently include only
cellular measurements) and so further
standardization and extension of mobility mechanisms is foreseen.
The way these measurements would
be handled by the UE is illustrated in
Figure 3, and the suggested trafficsteering process would work as follows:
1. the RAN instructs the UE by setting
thresholds or conditions on which UE
should perform measurement reporting.
For example, the RAN may provide a
condition based on signal strength –
RSSI – broadcasted WAN metric and
load or received power – RSRP;
2. if any of the thresholds are met, the UE
reports WLAN measurements back to
the RAN;
3. on the basis of this information, the
current network state and any other
information it has available, the
3GPP RAN will send a traffic-steering
message to the UE to switch data traffic
to or from WLAN.
Traffic steering today
– limitations
Today, the decision to move traffic
between 3GPP and Wi-Fi is made
independently by each UE based
on the UE’s implementation of the
connection manager. With varying
implementations, UEs from different vendors are likely to carry out the
switching decision at different times.
This makes the ecosystem unpredictable, increases capex and opex and
limits the operator’s ability to guarantee superior user experience.
The algorithm currently used by
UEs is oversimplified: the selection
of Wi-Fi over cellular is based solely on the availability of Wi-Fi. Once
a UE detects Wi-Fi, it automatically
shifts data traffic to it. Even if devices become smarter, UEs will still
make decisions based on a single cell
and a single AP – decisions that are
not coordinated with the network
(which can see all resources) or with
other UEs in the area. Most of the UEs
in the coverage area of an AP will
therefore try to connect to it.
sense multiple access (CSMA) protocol
degrades and the total capacity of the
AP drops rapidly.
To avoid this type of degradation,
an AP should not permit a UE to attach
unless the following conditions are met:
the service level provided to the UE over
Wi-Fi is better than the level offered over
cellular; and
the quality of the Wi-Fi connection is
good enough to ensure that the
experience level for users that are
already connected to the AP will not be
overly affected by the addition of
another user.
Proof of concept
By putting the network in control, and
using all the network information and
aggregated information from UEs in
the decision-making process, the network can provide the best user experience as well as increasing its overall
mobile broadband capacity. The proof
of concept measurements for such an
approach are shown in Figure 4.
The left-hand graph shows that by
rejecting users with bad link budgets
and poor radio conditions, the APs can
deliver higher throughput rates with
exactly the same equipment. Without
the mobility feature, the AP reaches a
maximum peak rate of 30Mbps, and
delivered throughput for most of the
UEs is between 0 and 17.5Mbps. With
the feature activated, the same AP
reaches a maximum peak rate of
What is Wi-Fi good at?
Let’s consider how Wi-Fi is designed:
to provide high peak rates and low
latency for a limited number of
users in the vicinity of an AP. When
too many users are connected, or
when the users have bad link budgets, the performance of the carrier
FIGURE 3 Connected mode mobility to and from Wi-Fi
3GPP RAN
The 3GPP mobility mechanisms for idle
mode will be extended in 3GPP ­Rel-12,
providing coordination between 3GPP
and Wi-Fi for UEs in idle mode.
While the suggested mobility feature
can evolve to include enhancements, it
can be included in any vendor solution
without requiring any direct interfaces,
thus increasing the level of coordination
between the various access networks,
cell types and carriers from different
vendors.
WLAN
2
BSS/WAN
3
1
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Switching the smart way
52
40Mbps, with most users receiving
between 20 and 40Mbps, and the capacity is increased up to 100 percent.
The right-hand graph shows that
without the mobility feature, UE performance drops by up to 20Mbps; 50
percent of users experience a drop of
10Mbps and all of them experience
inferior performance when moving to
Wi-Fi. With the feature activated, the
throughput difference in the worse case
is 7.5Mbps and on average the difference
is zero – thus mobile broadband experience is maintained.
Network probing
Another approach to optimizing connections is to force the UE to probe the
network before it switches. However,
this approach increases the uncertainty of measurements – the greater the
number of UEs probing the network,
the more unrealistic measurements
become. Probing can at best provide an
indication of the level of performance
available at a given moment, and results
cannot take into consideration changes
to network parameters created by SON
features, triggering of mobility features
or resource utilization by other UEs. The
lack of coordination among individual
FIGURE 4 UEs and the decisions made can lead to
devices ping-ponging from one access
technology to another, causing signaling storms and additional battery consumption – leading to a degradation in
overall network performance.
Many implementations
The variety of connection manager
implementations in UEs and the range
of radio sensitivities means that the outcome of today’s oversimplified mobility mechanism is unpredictable, which
makes it difficult for operators to plan
and optimize networks.
Measured capacity increase by RRM and coordination of Wi-Fi and 3GPP
C. D. F.
(%)
Legacy AP
Network controlled AP
100
90
80
60
60
50
40
30
20
10
0
2.5
Integration tools
Some complementary tools, such as
Hotspot 2.0 and ANDSF, have been developed to ease the integration of Wi-Fi.
By using HS2.0, broadcasted BSS load
and WAN metrics, UE-based switching
improves somewhat2, as these mechanisms can prevent UEs from connecting to overloaded or limited backhaul
Wi-Fi APs.
However, a low load does not necessarily imply that the resource availability of an AP is good. Low load can be the
consequence of a bad radio environment
caused by interference. To use these
mechanisms efficiently, the refresh
­frequency of broadcasted load update
needs to be fine-tuned to prevent mass
Approaches working against each other
Today, 3GPP mobility features are
deployed both within a radio access
technology and among technologies.
These features are controlled by the
RAN, and work in real time to provide
the desired mobile broadband experience and optimize resource management. They also take into account
network parameter changes, as SON
features are triggered to adapt network
behavior to constantly changing subscriber activity.
If the 3GPP/Wi-Fi decision continues
to be made by the UE, this can interfere
C. D. F.
(%)
0
with deployed 3GPP mobility mechanisms, causing devices to loop among
access networks and wasting resources. For example, an active 3GPP mobility feature might be in the process of
switching a UE to another technology
(from LTE to 3G, for example) when the
device decides to switch to Wi-Fi instead.
If the Wi-Fi becomes overloaded, it will
switch the UE back to the 3GPP network, which might trigger an additional switch to another cell.
5
7.5 10 12.5 15 17.5 20 22.5 25 27.5 30 32.5 35 37.5 40 42.5
100
90
80
60
60
50
40
30
20
10
0
-20.0 -17.5 -15.0 -12.5 -10.0 -7.5
Legacy AP
Network controlled AP
-5.0
-2.5
0
2.5
5.0
7.5
10
Throughput difference after selecting Wi-Fi
(Mbps)
Throughput in Wi-Fi
(Mbps)
Throughput
(%)
50
25
0
19.33.20
19.35.00
19.36.40
19.38.20
19.40.00
19.41.40
19.43.20
19.45.00
Time
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
19.46.40
19.48.20
19.50.00
19.51.40
19.53.20
19.55.00
19.56.40
53
toggling between cellular and Wi-Fi
accesses as a result of brief load peaks.
The ANDSF mechanism augments
PLMN selection, providing operators
with improved control over the decisions made by devices. Through a set
of static operator-defined rules, ANDSF
guides devices through the decisionmaking process of where, when and
how to choose a non-3GPP network. But
the final decision is still made by the
UE and is still dependent on the implementation of the device’s connection
manager. ANDSF is defined to take into
account the local environment – in other words, the specific connection manager implementation in the UE – and so
UEs from different vendors behave in
different ways. Using ANDSF might lead
to an unpredictable outcome.
Even if they are enhanced with radio
information, NAS-based solutions do
not have the real-time information
needed to make the optimal decision.
This is because requirements on information updates would become too
demanding for terminals and networks.
The ANDSF mechanism is, however,
valuable as it provides operators with
some level of control over the 3GPP/
Wi-Fi selection process, and it takes
into account static or slowly changing
network parameters. However, this
mechanism has an impact on battery
utilization in UEs, as it scans continuously for Wi-Fi and does not cater for
the rapidly fluctuating RF environment
experienced by a mobile device. Neither
does the mechanism take into account
UE and radio access capabilities (such as
LTE, HSPA, EDGE, carrier aggregation,
cell bandwidth, 802.11n and 802.11ac),
or network changes (as a result of selfoptimizing features in RAN). And it does
not provide real-time coordination with
decisions in other UEs in the area.
By deploying an HS2.0- and/or ANDSFonly approach, the same policy will be
applied by all UEs configured with that
policy, independently of their RF conditions, cell and network capabilities in
a manner that is not coordinated with
other UE decisions or network features.
This results in an inability to guarantee
the best user experience as well as creating additional opex and capex for the
operator, as resource utilization is not
optimized on a network level.
FIGURE 5 Usage of Wi-Fi shown by authentication method
SIM sessions
Web sessions
Jan
2011
Mar
2011
May
2011
Jul
2011
Sep
2011
Scenario: many users on a train
Consider this scenario: a train is
entering a station where two APs are
deployed. There are 200 subscribers on
the train with active mobile broadband
sessions.
The APs in the station perform best
when the number of UEs attached is less
than 20. If AP1 has a slightly better signal strength than AP2, it is possible that
180 of the UEs on the train will try to
connect to AP1, while the remaining 20
UEs will select AP2. This is because each
UE is ignorant of the decisions made
by all the other UEs. Subsequently, a
number of UEs connected to AP1 will
be steered to AP2 or back to 3GPP as a
result of the high load on AP1.
When two UEs apply the same policy, the one that supports carrier
aggregation should move data traffic
to Wi-Fi much later – which requires
coordination.
So while ANDSF can be used as a simple mechanism for offloading, it neither
guarantees the best user experience nor
optimal resource utilization.
Seamless authentication and
increased security
To secure the air link between a UE and
a hotspot, Passpoint devices use the
WPA2 Enterprise security protocol. This
is a four-way handshaking protocol that
is based on AES encryption, and offers
a level of security that is comparable to
cellular networks.
Nov
2011
Jan
2012
Mar
2012
May
2012
Jul
2012
The Hotspot 2.0 specification supports four commonly deployed standard protocols:
SIM-based authentication – EAP-SIM for
devices with SIM credentials and EAPAKA for devices with USIM credentials;
and
non-SIM-based authentication – EAPTLS for client and server-side
authentication, with a trusted root
certificate, and EAP-TTLS with
MS-CHAPv2 for user-name-password
authentication.
The Hotspot 2.0 specification complements WPA2 Enterprise security by
incorporating features, such as layer-2
traffic inspection and filtering, as well
as broadcast/multicast control, which
are often used to mitigate common
attacks on public Wi-Fi deployments.
Seamless authentication (where
users are not required to introduce a
user name and password) and increased
security are key to Wi-Fi usage. This
phenomenon is illustrated in Figure 5,
which shows the uptake of Wi-Fi in an
airport network enabled with EAP-SIM/
AKA authentication.
By reusing SIM mechanisms, subscribers only need to be provisioned
once in the HSS/HLR to use an operator’s mobile broadband.
Core integration and IP session
mobility in 3GPP/Wi-Fi
Connection to the mobile core networks from Wi-Fi equipment
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
Switching the smart way
54
gives subscribers direct access to operator services, and provides operators
with a mechanism to improve visibility, reuse their mobile broadband infrastructure, and integrate with different
systems (LI, DPI, session establishment
and management, policy decision and
enforcement, reuse of wholesale and
roaming agreements, access to operator
branded or hosted services, and online
and offline charging). Such an architecture provides a harmonized platform
for handling cellular and Wi-Fi access,
allowing Wi-Fi to become a continuation of the operator’s mobile broadband
experience.
Packet-core network integration
helps operators to gain control of noncellular traf­fic, and consequently of the
user experience.
Seamless handover – ­uninterrupted
IP session mobility between Wi-Fi
and 3GPP – is a key technology when
it comes to providing good user experience and mobility. To implement it,
Ericsson has taken a leading role in both
the standardization process and in product development.
In 3GPP Rel-11, the GPRS Tunneling
Protocol (GTP) – which is widely
deployed in mobile networks – was specified for the S2a interface. Use of this
interface, for trusted non-3GPP access,
makes it possible for UEs to connect to
the Wi-Fi network and utilize packetcore network services without mobility
or tunneling support in the UE. The S2b
interface was also included in the standard for connection to the core through
untrusted Wi-Fi.
3GPP Rel-12 standards will enable
GTPv2-based IP session mobility
between Wi-Fi and 3GPP over the
S2a interface. Ericsson together with
Qualcomm demonstrated how this
works at MWC 2013, and IP session
mobility is expected to appear in vendor products in 2014.
The Rel-12 solution will bring the
user experience even closer to mobile
E R I C S S O N R E V I E W • 90TH ANNIVERSARY • 2014
broadband by overcoming the limitations of two IP stacks (one for cellular and one for Wi-Fi ) used in current
UE implementation. The double stack
approach results in IP address reallocation every time devices move between
cellular and Wi-Fi.
Some applications have tried to
overcome these limitations by deploying tunneling mechanisms and using
buffers, so that users do not notice the
delay introduced in the session by the IP
address reallocation (reestablishment
of TCP sessions, socket reallocation,
packet loss and rerouting delays). But
these solutions are not optimal for realtime applications such as video streaming or voice calls; nor do they support
IP-address-dependent security mechanisms such as VPN services, and they
are not supported by enough UEs to be
mass-market. By moving the support
for IP address continuity to the modem,
it is possible for a device to keep its IP
address at handover in a way that is completely transparent to applications. This
is what 3GPP Rel-12 offers.
Conclusion
Next generation carrier Wi-Fi addresses the technical challenges relating to
mobile broadband Wi-Fi. By enabling
operators to add Wi-Fi capacity where
3GPP spectrum is scarce, Wi-Fi-based
business models can be included into
operator offerings to maintain optimal
mobile broadband experience.
With operators in control and Wi-Fi
integrated into heterogeneous networks, the mobile broadband experience will become harmonized and
provide users with the best possible
performance. By taking control, operators will be able to predict and monitor
the choice of connectivity, maintaining
the user experience and optimizing network resource utilization.
Ericsson will develop its 3GPP and
Wi-Fi portfolios based on the concepts
outlined above.
Ruth Guerra
joined Ericsson in 1999
and is currently strategic
product manager for Wi-Fi
integration at Product Line
Wi-Fi and Mobile Enterprise. She is an
expert in Wi-Fi and 3GPP networks,
focusing on the strategic evolution of
Wi-Fi and 3GPP RAN integration. She
holds an M.Sc. in Telecommunications
from the Technical University of Madrid
(Universidad Politécnica de Madrid),
Spain.
References
1. Infonetics Research, May 2013,
Carrier WiFi Offload and Hotspot
Strategies and Vendor Leadership:
Global Service Provider Survey,
available at: http://www.infonetics.
com/pr/2013/WiFi-Offloadand-Hotspot-Strategies-SurveyHighlights.asp
2. Ericsson, December 2012, Ericsson
Review, Achieving carrier-grade WiFi in the 3GPP world, available at:
http://www.ericsson.com/
news/121205-er-seamlesswi-fi-roaming_244159017_c?i
dx=10&categoryFilter=ericss
on_review_1270673222_c
Ericsson
SE-164 83 Stockholm, Sweden
Phone: + 46 10 719 0000
ISSN 0014-0171
297 23-3220 | Uen
Edita Bobergs, Stockholm
© Ericsson AB 2014