CEPT
ECC
Electronic Communications Committee
TG6(14)070rev2
Task Group 6
ECC TG6 – M6
Lyon, 13 – 15 May
Date issued:
13 May 2014
Source:
Arqiva
The relationship between ISD and SINR in LTE eMBMS networks for
fixed roof top reception
Subject:
Password protection required? (Y/N)
N
Summary:
To better understand the capabilities of LTE eMBMS and LTE Broadcast networks 1 for providing wide area
broadcast coverage for fixed reception, the relationship between the inter site distance (ISD) and signal to
interference plus noise ratio (SINR) has been studied for a range of cyclic prefixes (CP). The study was
based on a Monte Carlo analysis of the distribution of SINR across a cell embedded within a regular
hexagonal cellular network. For a rural environment, three cases were considered, a cell embedded within
a network, a cell at the edge of a network adjacent to another region and a cell at the edge of a network
adjacent to two different regions.
The results of the study indicate that for a cell embedded within a network the SINR achievable is a
function of the ratio of the ISD to the CP. For the scenarios considered the Monte Carlo simulations
indicate that a SINR of 16 dB or better is available at 95% of locations 2 when the ISD is ~20% of the CP for
an ISD of 1 km and ~50% of the CP when the ISD is 10km.
For cells located at the edge of a network adjacent to one other region an SINR of 6 dB is available at 95%
of locations. When adjacent to two other regions an SINR of 4 dB is available at 95% of locations. For the
cases considered a separation of between 1 and 2 ISD is required between cells in a border region before
the SINR matches that of an embedded cell. It should be noted that for a cell radius of 10 km this leads to a
zone along a regional or national boundary with reduced or no service of between 17 and 34 km in width.
The study shows that for a frequency re-use of 1, if universal coverage is required (95% of locations), the
SINR will be limited to 4 dB by edge cells. This corresponds to a data rate of approximately 0.8
bits/second/Hz. At such a data rate to replicate the ~240 Mbit/s available from six DVB-T2 multiplexes, an
LTE eMBMS network3 would require ~960 MHz of spectrum and a LTE Broadcast network ~300 MHz of
spectrum. If the coverage criterion is relaxed along regional and national boundaries then, within an area
away from the edge of a region an SINR of ~16 dB can be achieved which corresponds to a data rate of
approximately 2.5 bits/second/Hz.
Consequently for an LTE Broadcast network to replicate the coverage of existing terrestrial television
networks the following observations can be made;



1
If a frequency re-use of one is used and capacity is maintained then coverage will be compromised
or;
If a frequency re-use of one is used and coverage is to be maintained then capacity will be
compromised or;
If capacity and coverage are maintained, more spectrum, a frequency re-use of 3 or 4 will be
required.
In the context of this report LTE eMBMS refers to the broadcast capabilities of LTE as specified up to
within the 3GPP standard. LTE Broadcast refers to the proposals made by Qualcomm for an extended
cyclic prefix and dedicated downlink channel.
Proposal:
Consideration of alternative technologies and network topologies for terrestrial television
broadcast services must consider broadcaster coverage requirements. When specifying
assumptions, such as a frequency re-use of 1, the consequences of such an assumption must be
clearly stated.
Background:
ECC-TG6 should provide technical studies related to The evolution of broadcasting and mobile
networks and services as well as other services and applications.
In this context, broadcasting should encompass foreseen developments in video resolution,
coding, modulation/systems, receiving modes and coverage requirements. Mobile services
should include categories of data traffic, traffic asymmetry, network topologies, off-loading,
technologies such as eMBMS and Tower-Overlay. In addition, the concept of convergence/cooperation of both types of services/networks should be addressed.
2
Broadcaster service criterion for coverage is 95% locations at the edge of a service area not 95% of
locations within an area. An availability of 95% of locations at the edge of a service area corresponds to
an availability of more than 99% across an area. As such a relaxed service criterion has been used for
this study.
3
For the spectrum requirement calculation, considers the 60% downlink limit and limited use of the uplink
capacity during broadcast sessions.
2
Introduction:
One possible future scenario for terrestrial broadcasting is to move the platform from a high
tower high power HTHP) infrastructure to low tower low power (LTLP) based on existing cellular
infrastructure. The stated advantage of this, if coupled with a move to an LTE based solution, is
convergence between mobile and broadcast platforms plus potential release of spectrum. The
spectrum release is based on the claim4 that a LTLP network can operate with a frequency
re-use of 1 and still provide equivalent coverage to existing HTHP terrestrial television networks
that operate with a frequency re-use of between 5 and 6. This note presents results of initial
work aimed at investigating what the implications of a LTLP network operating with a frequency
re-use of 1 will be. In particular the study has looked at whether contiguous coverage can be
maintained in the area between two or more adjacent LTLP networks carrying different content.
Methodology:
Monte Carlo simulations have been carried out to assess the distribution of SINR across a cell
site located in various positions in a cellular network. The modelled network of cell sites was
composed of 14 rings of cell sites around a central site, a total of 631 cell sites. Depending on
the scenario being modelled, cell sites were designated either as being part of the same SFN as
the central site or as part of a different SFN, i.e. an interferer. Whether a cell site that was part
of the central site SFN was a contributor at a point depended on the position of the reception
point and the length of the cyclic prefix. For signal summation at the receiver a ‘brick wall’
approach has been assumed for the contribution of signals. Signals arriving within the cyclic
prefix were total contributors whilst those signals arriving outside the cyclic prefix are treated as
interferers. For determining which signals contribute and which interfere the receiver was locked
to the closest signal source, i.e. the cell site serving the cell. Parameters modelled are listed in
Table 1.
Propagation Model
Environment
Location Variation
Hata–Davidson
Open
5.5 dB
DTT Receive Antenna Pattern
DTT Receive Height
DTT Receive antenna pointing
ITU-R 419
10 m a.g.l
At Cell Site
Frequency
Base Station Antenna HRP
Base Station Antenna VRP
Beam tilt
Base Station Antenna Height
Base Station Location
Base Station EiRP
Network
540 MHz
Omnidirectional
ITU-R 1336
1°
30 m a.g.l.
Centre of Cell
64 dBm
14 layers (rings) of Cell sites around central site
(631 cell sites)
Trials per Simulation
Cyclic Prefix (variable)
Cell radius (variable)
10,000
4.7, 16.6, 33.3, 100, 200 µs
1, 2, 5, 10 km
Coverage requirement
95% of locations
Table 1: Parameters used in Monte Carlo simulations.
4
TG6(14) Doc048 ‘Simulation Results for Cellular-based broadcast’, Qualcomm TG6 Meeting #4
3
Results:
The distribution of the calculated SINR, for a cell embedded in a network, for cell radius of 1, 2,
5 and 10 km for different cyclic prefix are shown in Annex 1 in Figures A1-1 to A1-4. The effect
on the SINR across a cell in a border region for cells with radius 10 km and a CP of 100 µs is
shown in Figures A1-5 and A1-7.
Discussion:
For the cell size and EiRP considered in the study, the SINR for cells located away from a
border is limited by self-interference. The self-interference is a function of the ratio of the cell
site spacing (the inter-site distance ISD) and the cyclic prefix (CP) length. For smaller cells to
achieve an SINR of 16 dB or better at 95% of locations the ISD should be less than 20% of the
CP. As cell size increase the ratio of ISD to CP increases and, for a network composed of cells
with radius of10km cell, is approximately 50%.
For the case of 10km cell radius and a CP of 100 µs, in cells immediately adjacent to a border
the SINR is reduced to between 4 dB and 10 dB, Figure 1a. In cells not adjacent to the border
an SINR of 16 dB or better is achieved. The area with reduced SINR is two cells wide, one cell
in each region; a total width of ~34 km. Where the SFN is large, for example national and the
country is large such as France, the border area with reduced service represents a small
percentage of the total area. However, for smaller SFN, for example regional or for a smaller
country, the border area with reduced service can represent a significant proportion of the
service area. It should be noted that in the border area a proportion of the area will still receive a
SINR of 16 dB or better, Figure 1b.
16dB
16dB
16dB
16dB
16dB
16dB
10dB
16dB
10dB
10dB
6dB
10dB
6dB
6dB
6dB
16dB
4dB
10dB
10dB
7dB
16dB
16dB
7dB
16dB
16dB
16dB
7dB
16dB
7dB
16dB
16dB
7dB
95%
95%
95%
95%
95%
78%
66%
78%
66%
95%
95%
95%
66%
95%
95%
95%
66%
95%
Figure 1a: SINR distribution in border area
cells : Cell radius 10 km
95%
52%
59%
95%
95%
78%
59%
59%
59%
95%
78%
78%
78%
66%
Figure 1b: Percentage of locations in border
area cells where the SINR is 16 dB or better :
Cell radius 10 km
Note: two regions denoted by blue and red border
hexagons
The SINR, based on the information in Table A2-1 in Annex 2, can be converted to an available
data rate Table 1.
SINR
Bit Rate per Hz
Data Rate in 10 MHz channel
4
~0.8
~8 Mb/s
16
~2.5
~25 Mb/s
Table 1: LTE Data Rate in 10 MHz channel
Based on the LTE data rate available the amount of spectrum required to replicate the capacity
of existing terrestrial television services can be estimated. Assuming terrestrial television is
based on 6 DVB-T2 multiplexes delivering 240 Mb/s, then
4
A LTE Broadcast network based on 10km radius cells using a 100 µs cyclic prefix would require
96 MHz to deliver 240Mb/s. Such a network, with a frequency re-use of 1, would have areas
along border regions that have reduced or no service.

To provide universal coverage such a LTE Broadcast network would require additional
spectrum in border areas. Typically a frequency re-use of 3 meaning that 3 x 96 MHz of
spectrum (288 MHz) is required, or;

To provide universal coverage with a frequency re-use of 1 a LTE Broadcast network
would require 300 MHz (240 Mb/s / 0.8 b/s/Hz)

A LTE eMBMS network because of a 60% limit on downlink only and the need to be
paired with an uplink that would be largely unused during broadcast sessions, would
require 320 MHz. As with LTE Broadcast, such a network, with a frequency re-use of 1,
would have areas along border regions that have reduced or no service.

To provide universal coverage a LTE eMBMS network would require additional spectrum
in border areas. Typically a frequency re-use of 3 meaning that 3 x 320 MHz of spectrum
(960 MHz) is required, or;

To provide universal coverage with a frequency re-use of 1 a LTE eMBMS network
would require 1000 MHz.
The results are based on the directional receive antenna pointing at the cell site serving the cell.
Such a strategy is not optimal and a better SINR could probably be obtained by pointing in the
direction that provides the best SINR. Though, it must be noted, such an approach may be
difficult to implement in practice; the best receive antenna direction having to be calculated for
each location. Whilst the SINR could be improved by adopting a more optimal solution it must
also be recognised that if time variability5 of signals were to be included, self-interference and
interference from adjoining regions would be worse than calculated. Further work is required to
investigate the effect of both an optimal pointing strategy and the impact of time variability of
signals.
The study only considered cellular networks composed of cell sites of the same radius, and with
transmitters of uniform effective height. Mixed networks should also be modelled; that is
networks consisting of clusters of sites at say 1 km radius representing urban areas, surrounded
by a ring or two of 2km sites representing suburban areas, or just clusters of 2km suburban cells
within a 10km radius cellular network. This is again an area that requires further study, as well
as networks with transmitters of variable effective height.
Conclusion:
For the scenarios considered the results of the Monte Carlo simulations carried out indicate that
in border areas between LTLP networks carrying different content, i.e. are not part of the same
SFN, reduced coverage can be expected.
Based on the assumptions in this study, to provide a service in these border areas either a
reduced data rate must be accepted if a frequency re-use of 1 is adopted or, to serve the areas,
a frequency re-use of at least 3 will be required. In either case providing a service in border
areas leads to a higher spectrum requirement.
If special measures are not taken to serve border areas, coverage in such areas will be
reduced.
Further work is still required to better establish the extent of reduced coverage in the border
areas between two different LTLP networks.
5
Whilst time variability of signals is not normally accounted for in coverage predictions of unicast cellular
networks, it is a feature of broadcast predictions. Unicast networks can adapt to variable propagation
conditions by varying the data rate, such measures are not available to broadcast networks that need to
maintain a constant data rate and quality of service across the entire coverage area.
5
Annex 1: Simulation Results
Figures A1-1 to A1-4 show the distribution of SINR in a cell embedded in the centre of a cellular
network, i.e. a cell away from the edge. Results are for cellular networks composed of cells of 4
different radii with the cyclic prefix for each radius varied between 4.7 and 200 µs. For the 1 km
cell radius case the 200 µs case has not been plotted as it overlay the 100 µs case.
Figures A1-5 to A1-7 show the distribution of SINR in a cell close to the border between SFN
carrying different content; cells coloured red/purple are part of the wanted SFN and cells in blue
areas the unwanted SFN. Individual dots mark the cell sites which are located at the centre of
hexagonal cells. The white cell site is the cell site within which the SINR is calculated.
6
Figure A1-1: SINR distribution in a 1km embedded cell
Effect on SINR of varying the cyclic prefix : 1km Cell radius
100.00%
90.00%
80.00%
70.00%
4.7 uS
16.6 uS
33.3 uS
100 uS
CDF
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
0
-10
10
20
40
30
80
70
60
50
SINR dB
Figure A1-2: SINR distribution in a 2km embedded cell
Effect on SINR of varying the cyclic prefix : 2km Cell radius
100.00%
90.00%
80.00%
70.00%
4.7 uS
16.6 uS
33.3 uS
100 uS
200 uS
CDF
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
-10
0
10
20
30
40
SINR dB
7
50
60
70
80
Figure A1-3: SINR distribution in a 5km embedded cell
Effect on SINR of varying the cyclic prefix : 5km Cell radius
100.00%
90.00%
80.00%
70.00%
CDF
60.00%
4.7 uS
16.6 uS
33.3 uS
100 uS
200 uS
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
-10
0
10
20
40
30
60
50
70
80
SINR dB
Figure A1-4: SINR distribution in a 10km embedded cell
Effect on SINR of varying the cyclic prefix : 10km Cell radius
100.00%
90.00%
80.00%
70.00%
CDF
60.00%
4.7 uS
16.6 uS
33.3 uS
100 uS
200 uS
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
-10
0
10
20
30
40
SINR dB
8
50
60
70
80
Figure A1-5: SINR distribution in a 10km cell at a North-South region boundary
Effect of one adjacent region on SINR : 10km Cell radius : 100uS cyclic prefix
100.00%
90.00%
80.00%
70.00%
0 ISD
1 ISD
2 ISD
3 ISD
CDF
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
0
10
20
30
40
50
SINR dB
Region separation 1 ISD
Region separation 0 ISD
Region separation 3 ISD
Region separation 2 ISD
9
60
Figure A1-6: SINR distribution in a 10km cell at an East-West region boundary
Effect of one adjacent region on SINR : 10km Cell radius : 100uS cyclic prefix
100.00%
90.00%
80.00%
70.00%
0 ISD
60.00%
CDF
1 ISD
50.00%
2 ISD
40.00%
30.00%
20.00%
10.00%
0.00%
0
10
30
20
40
50
SINR dB
Region separation 0 ISD
Region separation 1 ISD
Region separation 2 ISD
10
60
Figure A1-7: SINR distribution in a 10km cell at two region boundaries
Effect of two adjacent regions on SINR : 10km Cell radius : 100uS cyclic prefix
100.00%
90.00%
80.00%
70.00%
CDF
60.00%
0 ISD
1 ISD
2 ISD
3 ISD
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
0
10
20
30
40
50
SINR dB
Region separation 0 ISD
Region separation 1 ISD
Region separation 2 ISD
Region separation 3 ISD
11
60
Annex 2: LTE Data Rate
Spectral Efficiency
b/s/Hz
C/N Gaussian C/N Ricean
dB
dB*
0.14
-5.00
-2.50
0.18
0.22
0.29
0.36
0.44
0.52
0.62
0.70
0.80
0.80
0.88
0.99
1.14
1.30
1.41
1.53
1.53
1.64
1.83
1.98
2.14
2.29
2.55
2.74
2.83
3.06
3.17
-3.75
-2.50
-1.88
-0.63
0.31
1.25
2.50
3.75
4.06
4.38
5.00
6.25
7.50
8.13
9.06
9.38
10.00
11.25
12.50
13.44
14.06
14.38
15.31
16.88
17.50
19.38
20.00
-1.25
0.00
0.63
1.88
2.81
3.75
5.00
6.25
6.56
6.88
7.50
8.75
10.00
10.63
11.56
11.88
12.50
13.75
15.00
15.94
16.56
16.88
17.81
19.38
20.00
21.88
22.50
*Gaussian + 2.5 dB
Table A2-1: LTE Data Rate
12