Version 1 25 February 2009
LTE/SAE Trial Initiative
Latest Results from the LSTI, Feb 2009
Julius Robson, Nortel
Chairman LSTI Proof of Concept Group
1
www.lstiforum.org
Version 1 25 February 2009
Contents
• LSTI’s Objectives
• Who’s involved?
• LSTI Activities
• Latest results from the Proof of Concept Group
• Inter-Operability Testing
• Friendly Customer Trials
• Activity timelines
• Summary
2
Version 1 25 February 2009
The LTE /SAE Trial Initiative
• The LSTI is an open initiative driven by Vendors
and Operators launched in May 2007
• Its objectives are to:
• Drive industrialization of 3GPP LTE/SAE technology
• Demonstrate LTE/SAE capabilities against 3GPP
and NGMN requirements
• Stimulate development of the LTE/SAE ecosystem
3
Version 1 25 February 2009
LSTI Participants
23 LTE Equipment Vendors
9 Operators
4
…representatives from across LTE’s Global Ecosystem
Version 1 25 February 2009
LSTI Activities
From Standardisation…
Proof of Concept
Can the design be implemented? Are the performance targets achievable?
Interoperability
Does everyone have the same interpretation of the standard?
Friendly Customer Trials
What can it deliver in near commercial conditions?
…to Commercial Rollout
5
Version 1 25 February 2009
Latest Results
From the Proof of Concept Activity
Demonstrating that basic LTE/SAE functionality and performance
are achievable with pre-standards proprietary equipment
6
Version 1 25 February 2009
“Proving The Concept” of LTE/SAE
Peak
data rates
Radio
latency
Multi Cell Live Air
Multi User End-End
Lab tests
Vehicle
speeds
MIMO
gains
Multi user
scheduling
Connection
Setup time
QoS
VoIP
Real world
Data rates
Handover
Early testing focused on fundamental performance in ideal conditions…
7
…later tests required end to end architecture and real world scenarios
Version 1 25 February 2009
Proved in
Proofpoint Status
Each filled circle represents one set of
vendors tests which demonstrate the
proofpoint is feasible
Proved In
Observed: LSTI Operators visited the
vendor’s test facilities and observed
successful demonstration of the proofpoint
A status indicator shows industry progress for each proofpoint
8
Version 1 25 February 2009
PoC Results
Part 1) Data Rates
How much will you get?
LTE is designed to deliver over 320 Mbps throughput
…But what can be achieved in practice?
And what data rates will users actually experience?
Part 2) Latency
How quickly will you get it?
9
Version 1 25 February 2009
Proved in
LTE working in the field
LTE has been proved working in the field both in single cell and multi cell scenarios
Single Cell
Multi Cell
-- Single Stream
-- Dual Stream / MIMO
Handover at speeds up to 100km/h
MIMO working in the field
50
Tput Mbps
FTP
restart
40
Mbps
246Mbps on a
drive test
with 4x4 MIMO
30
20
10
HO
HO
0
10
48Mbps Tput on
multicell drive test
2x2 MIMO
10MHz BW
0
60
120 180 240 300 360 420 480
sec
Version 1 25 February 2009
Proved in
Peak Data Rates & Spectral Efficiency
LTE-FDD
TD-LTE
9
350
300
L1 pk rate,
Mbps
Mbps
250
200
DL 2x2
8.6bps/Hz
150
100
UL 64QAM
UL 16QAM
50
2.8bps/Hz
bps/Hz
L1 Spectral Efficiency,
bps/Hz
DL 4x4
DL 2x2
Lab results
Field results
DL target
UL target
8
7
6
5
UL 64QAM
4
3
UL 16QAM
2
1
0
0
0
0.2
0.4
0.6
0.8
1
Code rate
0
0.2
0.4
0.6
0.8
1
Code rate
results normalized 3GPP overheads & 20MHz bandwidth
Peak rates are the top speed of the system
…achieved in optimal signal conditions with a single user in the cell
Requirements are 100Mbps or 5bps/Hz for DL, and 50Mbps or 2.5bps/Hz for UL
Measured Peak rates in lab and field meet the requirements
11
Version 1 25 February 2009
…So Will All Users Experience >100Mbps?
• Peak rate requirements apply to corner point conditions
which can be verified by simulation or lab testing
• Actual rates that users experience will be impacted by:
Protocol Layer
Pk Rates Represent
Corner Conditions
•One UE/cell
•Optimal RF
•Phy layer
ll
ce
/
s
UE
Path Loss / UE Speed
1) RF conditions & UE speed
Peak rates represent optimal conditions,
lower rates are experienced towards the cell edge
and when the UE is moving at high speed
2) Multiple users in the cell
UE data rates will be lower when sharing the cell with
others
3) Application Overheads
Peak rate requirements apply to Physical layer.
There will be overheads when considering data
transfer between applications
• Impacts to UE rates are analysed in the following slides….
12
Version 1 25 February 2009
Radio Conditions – Signal quality
Tput vs SNR typical example DL result
Tput relative to peak
100%
80%
60%
40%
20%
0%
5
Cell edge
10
15
SNR, dB
20
25
Near base
Example results shown,
similar behaviour observed
for SM MIMO, SIMO and
SFBC and UL SIMO
Peak rates are achieved with high signal quality near the base station
Tput is lower towards the cell edge
13
Version 1 25 February 2009
Proved-in
Radio Conditions – UE Speed
Tput vs SNR typical example DL result
Tput relative to peak
100%
80%
60%
40%
3 kmh
30 kmh
120 kmh
240 kmh
350 kmh
20%
0%
5
Cell edge
10
15
SNR, dB
20
25
Near base
Example results shown,
similar behaviour observed
for SM MIMO, SIMO and
SFBC and UL SIMO
Resiliency of LTE prototypes to high user speeds is tested in the lab
Initial results demonstrate support of up to 350km/h
Little impact to throughput is seen at speeds up to 120km/h
14
Version 1 25 February 2009
Proved in
Predicted End User Data Rates
End user Tput, Mbps
ll
ce
gle
sin
15
,
ak
Pe
20
d
de
loa
..from the best case
Peak rate in an isolated cell
25
%,
p5
To
The green area is an
extrapolation showing the
variation of user data rates…
for 20MHz bandwidth
e
ag
er
Av
Consolidated measurements
show cell edge and average
user Tputs with ‘busy hour’
interference levels
and 1 UE/cell,
10
5
Bottom 5%
ave
cell edge
0
1
..to the worst case
Bottom 5%-ile in busy hour
2
3
4
5 6 7 8 9 10
20
Active Downloads per cell
LSTI Measurements
(1 UE / cell)
Extrapolation assuming
a fair scheduler
NGMN Targets
(10 UEs/cell)
Predicted end-user Tputs based on LSTI lab measurements
align with NGMN Targets for ‘cell edge’ and average Tput
15
Version 1 25 February 2009
Proved-in
Multiple Users per Cell - Downlink
Sharing of Downlink Tput
L1 Tput, Mbps
At any given instant, the
cell’s spectral resource
is shared between all
active users
Lab test with flat AWGN channels
Idle UEs
Active UEs
pwr
Active
UE1
UE2
frequency
Cell Tput Mbps
Frequency Selective Scheduling
2UEs
1UE
20
10
0
0
10
20
SNR (dB)
The scheduler can exploit different frequency responses
of each UE’s channel, to increase cell Tput
16
30
Version 1 25 February 2009
Multiple Users per Cell - Uplink
Cell Tput depends on
the mix of RF conditions
for the active UEs
Proved-in
Mbps
Multi UE Uplink Tput
with various mixes of RF conditions
Good
Med
Good
Med
Med
Good
Poor
Poor
Poor
Sharing of Uplink Throughput during a drive test
100%
Idle UEs
The LTE Uplink has Multi-User
MIMO, which pairs-up UEs to
share the same UL resource to
increase cell Tput
17
% Usage MU-MIMO
Active UEs
Paired-up
0%
Version 1 25 February 2009
Proved In
Throughput at the Application Layer
Over-the-air bits
Coding, Control,
HARQ Ref signals
L1- (PHY) Tput
MAC
MAC Tput
header
RLC PDCP, IP &
TCP/UDP headrs
TCP/UDP Tput
Application
Not to scale
header
Lab measurement (10 MHz, SIMO)
Application Tput
Field measurement (20 MHz, 2x2 MIMO)
30
100
TCP
Layer 1
90
25
80
70
60
CDF [%]
Mbps
20
15
50
40
10
L1 (AMC + HARQ)
Application Layer
5
30
L1
20
APP
10
0
0
10
20
SNR (dB)
30
0
0
20
40
60
80
100
Throughput [Mbps]
120
140
160
Throughput requirements are specified at L1 (Physical Layer)
Measurements show the difference between L1 & Application to be small
for large packets (e.g. File Transfer)
18
Note IP packet size affects the amount of protocol overhead
Version 1 25 February 2009
Proved in
Guaranteed Bit Rate Support
Bes
t
Guaranteed
Bit Rate pipe
effo
Best effort
rt
GBR
Cell edge
The GBR user is allocated resource to maintain their
desired rate. The best effort user is given the rest
Since end user data rates can vary with radio conditions and network loading,
LTE has to provide Guaranteed Bit Rate (GBR) links, in order to support
streaming services such as IPTV or internet radio
Several vendors have demonstrated that constant bit rates can be maintained
independently of signal quality and other traffic on the cell
19
Version 1 25 February 2009
PoC Results
Part 1) Data Rates
How much will you get?
Part 2) Latency
How quickly will you get it?
To provide an ‘always on’ experience,
LTE/SAE requires low delays for both
user data and control of resources
20
Stopwatch by P.Neal www.PNeal.com
Version 1 25 February 2009
Proved in
Control-Plane Latency: Idle to Active time
• To provide many users with an ‘always-on’
experience, LTE is designed with a low idle to
active transition time
• All UEs sit in an idle state when there is no data to
transfer – but can be activated quickly when they
need to communicate
Measured Idle-Active Times
3GPP target 100ms
Measured with one UE/cell
Idle UEs
Active UEs
active
idle
0
20
40
60
80
100
120 ms
Measured idle to active times meet the 100ms requirement
21
Version 1 25 February 2009
DRAFT
Proved
in
User-Plane Latency
EPC
Server
eNB
UE
ping
U-Plane: Measured Round Trip Times
3GPP target
(Air interface)
Air interface
End - End
0
5
10
Pre-scheduled uplink
15
20
25 ms
On demand uplink
Low user-Plane latency is essential for delivering interactive services, like gaming and VoIP
Measured round trip times with a pre-scheduled uplink meet the 3GPP target of 10ms
Unpredictable traffic requires an on-demand uplink, E2E round trips times are under 25ms
22
Version 1 25 February 2009
DRAFT
Proved
in
VoIP Support
eNB
ding
a
Best effort lo
EPC
Measured Performance in loaded conditions
Packet Latency
Jitter
(one way)
0,6
50
50
40
40
30
30
ms
ms
Packet Loss
0,5
0,4
%
VoIP
QoS
Server with
VoIP Test Tool
0,3
20
20
0,2
10
10
0,1
0
As with data rates, latency can be
impacted by signal conditions and loading
0
0
=acceptable performance
Tests in ideal lab conditions have demonstrated that LTE/SAE is capable of providing
IP connectivity with sufficient latency, jitter and packet loss performance to support
good quality VoIP, even in the presence of other traffic.
Operators observed good quality VoIP maintained in live air drive tests with handover
23
Version 1 25 February 2009
Proved in
Uplink Transmit Power Control
20
Normal Power Control
UE Tx Power (dBm)
10
UL Rx Power (dBm)
Power (dBm)
Three companies demonstrated open
and closed-loop TPC working in the lab
or field. Results shows TPC
maintaining desired UL received power
at eNB to compensate for path loss.
0
-10
-20
-30
-40
-50
-60
-70
30
40
50
Fractional Power Control
Target Rx power is low for
cell edge user to reduce
the interference to
neighbour cell
Path loss PL
0
Near cell site
80
90
Cell edge user
Fractional TPC
CTPC
1.5
Throughput (Mbps)
Target Rx power
24
Factor : -(1 – )
70
2
An initial result shows Fractional TPC increases
Tput near the base station with only a small
increases in interference to other cells
Po
60
Pathloss (dB)
1
0.5
0
70
80
90
100
110
Measured path loss (dB)
120
Version 1 25 February 2009
Proved in
Measurement Functionality
-60
CellID 184
-65
CellID 185
RSRP (dBm)
-70
-75
-80
-85
-90
-95
-100
17:23:19.783
Handover
17:23:22.662
17:23:25.652
17:23:28.773
17:23:31.652
time
The UE must perform measurements and report them back to the eNB, for
use in Radio Resource Management decisions.
The above example shows UE measurements of pilot signals from two cells.
The eNB performs a handover once the new cell is consistently stronger.
UE Measurements have also be shown to enable uplink power control,
frequency selective scheduling and link adaptation
25
Version 1 25 February 2009
Proved in
Handover
• Inter-eNB and intra-eNB handovers demonstrated
in the lab and field at up to 120 km/h
• Data interruption times under 50ms achieved,
meeting NGMN’s ‘real time service’ requirement
• Both S1 and X2-assisted handovers demonstrated
• X2 Improves handover performance and
reduces loading on MME
Intra eNB handover
EPC:
Gateway
& MME
Source
eNB
26
S1
S1
X2
Target
eNB
Inter eNB handover
Version 1 25 February 2009
Summary of PoC Results
• Over 80 sets of tests from 8 vendors consolidated into 15
proofpoints, demonstrating feasibility of key LTE/SAE
functionality and performance:
• 3GPP and NGMN targets for peak data rates and minimum
latency are achievable for both FDD and TDD variants
• ‘Real world’ performance results are also given, helping
operators understand what can be offered to end users
• QoS has been demonstrated to provide the consistent data
rates and latency needed to support services like VoIP
• Benefits of MIMO, Frequency Selective Scheduling and
fractional power control technologies are shown
• Demonstration of key functionality like Handover and UE
measurement reporting shows the maturity of developments
27
LTE/SAE does exactly what it says on the box
Version 1 25 February 2009
Interoperability Testing (IOT)
• Successful IOT shows that everybody has the same
interpretation of the standard, implying a level of maturity
• LSTI agrees and recommends a core set of features for
interoperability testing on the LTE/SAE interfaces
• First phase: IODT (D=Development)
• Focus on the Air interface: IOT between UEs and eNBs
• Feature set and standards baselines agreed
• Testing underway, reporting during Q2-Q4 2009
S1
• Second phase: IOT
• Extra features and multiple partners
for the Air Interface
• S1 and X2 testing, requiring multiple
RAN and EPC vendors
• Expected to complete mid 2010
28
EPC
S1
eNB
X2
Uu
eNB
Version 1 25 February 2009
Friendly Customer Trials
• The friendly user trial phase enables Vendors and Operators to
prepare for deployment and commercial launch
• Visualize LTE capabilities and advantages in nearly commercial conditions
with test applications provided by vendors and 3rd parties
• Status
• Trial test cases based on NGMN field trial requirements have been drafted
and are under peer review
• Content and timing of phases agreed:
1) Early testing of Radio access
systems – by end 2009
2) Integration of EPC to enable
End-to-end testing – by mid 2010
29
Version 1 25 February 2009
LSTI Activity Timing
2007
2008
2009
Proof of Concept
IODT
IOT
Friendly Customer Trials
PR/Marketing
30
preparation
preparation
preparation
2010
Version 1 25 February 2009
In Summary,
The LTE/SAE Trial Initiative….
…is an open initiative of vendors and operators working
together to accelerate the development of a global
ecosystem for LTE
…provides cross-industry co-ordination of prototyping,
interoperability testing and field trials
The Proof of Concept Activity…
…has shown that it is feasible to make LTE/SAE equipment
that can meet industry targets for peak performance
…is also revealing the performance that operators will be
able to offer to end users in real world conditions
31
Thank You
www.lstiforum.org
32