Professional Documents
Culture Documents
Mihaela Beluri, Erdem Bala, Yuying Dai, Rocco Di Girolamo, Martino Freda, Jean-Louis Gauvreau, Scott Laughlin,
Debashish Purkayastha, Athmane Touag
InterDigital Communications, LLC
I.
INTRODUCTION
DEPLOYMENT SCENARIO
BACKGROUND
and may also provide the location of its slave devices, if these
devices are also equipped with geo-location capability. Unlike
the case of the FCC, the WSDB provides a master device with
the maximum transmit EIRP for each of the TVWS channels
and a time of validity for this information. The master device
then communicates to the WSDB the channels and actual
transmit EIRP it will use. A master device is also responsible
for ensuring that its slave devices transmit within the limits
specified by the database. The Ofcom VNS also specifies the
maximum required ACLR characteristics for any device using
the TVWS, as well as additional information (technology
identifier, device identifier, antenna characteristics, etc) to be
provided to the WSDB before a maximum EIRP can be
calculated by the database for each channel. Although the VNS
and an eventual European Harmonized Standard (to be used
throughout the European union) will not specify the algorithm
for EIRP calculation to be used by the WSDB, it is expected
that each regulator/country (e.g. Ofcom for the UK) will
develop this algorithm based on factors such as the proximity
of the white space device to nearby TV broadcast systems, the
probability of interfering with these systems, and potentially
the effect of multiple white space devices transmitting in an
area.
DTX DTX
DL
CG
DTX DTX
LBT
LBT
LBT
LBT
LBT
LBT
Sub-frame boundary
(transmission opportunity)
DL
CG
DL
multiple
subframes
1 subframe
Coexistence
gap (multiple
subframes)
Figure 2.
B.
No Transmission
(channel busy)
DL
CG
TON
TOFF
CPP
Figure 3.
DL
CG
HeNB
transmits
HeNB
Silent (no Tx)
A)
LTE Transmission
pattern
DL
CG
TON
TOFF
CPP
DL
CG
HeNB
transmits
HeNB
Silent (no Tx)
Wi-Fi Transmission
B) Pattern (long packet,
defers to other SU)
Wi-Fi interfers with LTE
Wi-Fi Transmission
C) Pattern (short packet,
defers to other SU)
Wi-Fi interfers with LTE
D)
Wi-Fi Transmission
Pattern (short packet,
does not defer to other
SU)
Wi-Fi interfers with LTE
50% or less (see Section V.B). For low LTE loads, the duty
cycle may be set to a minimum value. Note that in cases where
Wi-Fi and LTE traffic loads are high, a duty cycle of 50% may
not be enough to accommodate the Wi-Fi traffic. In this case, if
the Wi-Fi system does not defer to LTE, it will create a high
level of interference. In response, the LTE system may switch
the channel to one of the candidate channels, as indicated in
Section II. The high level algorithm for dynamically adapting
the coexistence duty cycle is shown in Fig. 5.
The coexistence duty cycle adaptation algorithm may run
Start
Yes
Yes
Wi-Fi
detected?
No
No
Yes
LTE load=LO?
End
No
Simulation scenario #2
The goal for this set of simulations is to asses how the LBT
and the CG gap approaches used by LTE affect the Wi-Fi
performance. For this scenario, simulations were run for
various values of the coexistence duty cycle.
One difference with respect to the simulation presented in
Section V.A is that all Wi-Fi nodes can sense the LTE
transmissions and defer to it. Similarly, when LTE uses the
LBT method for coexistence, the LTE nodes sense the Wi-Fi
transmissions and do not access the channel.
Other relevant simulation settings are shown below:
No hidden nodes.
RTS/CTS is enabled in Wi-Fi nodes.
Wi-Fi MPDU (MAC layer payload) = 1500 Bytes.
AP
STA
Cellular transmitter
UE
50
0
0
-50
-50
50
100
150
-100
-150
The data rates for the Wi-Fi STA were set as a function
of the distance between the STA and the AP, to 54, 48,
36, 24, 18, 12, 9, 6 Mbps, corresponding to a distance
of 26, 34, 40, 51, 63, 77, 84, 100 meters, respectively.
Wi-Fi loads = 4 Mbps, and 9 Mbps.
50
-50
Wi-Fi t-put
6.3%
B.
150
-100
-100
18%
150
100
packet
LBT
50% LTE channel usage
ratio
-82 dBm LTE sensing
threshold
10 ms LTE Tx duration
A. Simulation scenario #1
The goal for this set of simulations is to asses how the LBT
and the CG gap approaches used by LTE affect both the Wi-Fi
throughput, and the LTE throughput. Scenario #1 is a high
interference scenario, where 4 Wi-Fi APs and 3 LTE eNBs are
randomly placed in a circle of radius 120m, and STA and UE
receivers are placed around the transmitters in a circle with
radius 60 m. The simulation uses 3 STA per AP, and 3 UEs per
eNB. The transmit power for the AP and eNB is set to 100
mW. Only distance dependent free space propagation path loss
has been used to model the channel (no fading or shadowing
were modeled). All nodes interfere with each other. The
channel access is managed by CSMA/CA for the Wi-Fi nodes,
and LBT or coexistence gaps by the LTE nodes. The Wi-Fi
nodes are configured to use a 20 dB higher energy detection
threshold for non-Wi-Fi systems, as compared to the preamble
detection threshold for Wi-Fi. The throughput for all nodes is
computed by using the Shannon capacity (no additional margin
was added) to be able to observe only the effect of channel
access strategies of the cellular nodes. However, different
SINR ranges have been used for 802.11 and cellular nodes, as
follows: [6 .. 23] dB for the Wi-Fi nodes and [-2 .. 23] dB for
the cellular nodes. The results are averaged over 20 topologies.
Fig. 6 illustrates two sample topologies.
100
CG
50% duty cycle
2 ms Wi-Fi
length
-100
-50
50
100
150
200
3.5
3
2.5
2
1.5
LBT
Coexistence Gap
9
0.5
10
20
30
40
50
60
70
80
Percentage of Time LTE occupies the channel (%)
90
100
6
5
C.
4
3
2
LBT
Coexistence Gap
1
0
10
20
30
40
50
60
70
80
Percentage of time LTE occupies the channel (%)
90
100
Case 2
Case 3
Case 4
LTE system
Coexistence Gap, channel usage
ratio set to 50%
Traffic load 11.2 Mbps = 50%
channel capacity (corresponds to
50% of the measured LTE
throughput for the baseline)
Full buffer = Yes
No coexistence gap
Traffic load = Same as Case 1
Full buffer = No
Coexistence Gap, channel usage
ratio set to 50%
Full buffer = Yes
No coexistence gap
Full buffer = Yes
Wi-Fi System
All Wi-Fi defer
No Wi-Fi defers
No Wi-Fi defers
B.
Figure 9. OPNET test bench layout
Simulation Results
The throughput of the LTE and Wi-Fi systems, measured in
the TVWS channel, for each of the cases defined in Table III, is
shown in Fig. 10.
Figure 10 shows that in Case 1 and Case 2 (All Wi-Fi
defer), there is no degradation of the measured LTE
throughput compared to the submitted load. This is expected,
as for these cases, the LTE transmissions were not interfered by
the Wi-Fi. In Case 1, when the LTE system uses a gap pattern
with a 50% duty cycle, the measured Wi-Fi throughput equals
the Wi-Fi submitted load. When the LTE system does not
allocate coexistence gaps to transmit the same load (i.e. the
LTE system does not operate full buffer), the Wi-Fi
T-put (Mbps)
10
8
6
LTE t-put
4
Wi-Fi t-put
2
0
Case 1
50% CG
LTE full buffer
All Wi-Fi defer
Case 2
No CG
LTE not full buffer
All Wi-Fi defer
Case 3
50% CG
LTE full buffer
No Wi-Fi defers
Case 4
No CG
LTE full buffer
No Wi-Fi defers
Figure 10. Throughput of LTE and Wi-Fi systems on the TVWS channel
support LBT in LTE, and the fact that the operation in highly
congested channels is to be avoided, we feel that enhancing
LTE to support LBT is not justified. Moreover, the simulation
results show that the coexistence gap is a feasible approach to
non-coordinated coexistence with other secondary users of the
TVWS channels. Lastly, the simulations for different static gap
durations presented in Section V.B, suggest that extending or
reducing the gap length as a function of the traffic and
interference as explained in Section V.C, is a promising
method for friendly non-coordinated coexistence with other
SUs.
Future work includes extending the simulations to the
dynamic adaptation of the coexistence pattern parameters, as
well as investigating the performance of the coexistence gap
approach, to LTE systems operating in TDD mode.
REFERENCES
[1]
[2]
[3]
VII. CONCLUSIONS
In this paper, we introduced the Coexistence Gap (CG)
based approach to non-coordinated coexistence and evaluated it
by comparing to the Listen Before Talk (LBT) based
approach. In addition, the paper presented simulation results for
the comparison of the two methods, as well as for the
investigation of optimal coexistence pattern parameters.
The simulation results show that performing LBT in LTE
results in less degradation of the Wi-Fi throughput, in highly
congested channels, as compared to the non-LBT method
analyzed in the paper. However, with realistic implementations
of the LBT, this benefit may be smaller than indicated by the
ideal simulations. Given the extent of changes required to
[4]
[5]
[6]
[7]
[8]
http://www.attinnovationspace.com/innovation/story/a7781181
http://money.cnn.com/2012/02/22/technology/wireless_carrier_mergers/i
ndex.htm
IEEE P802.11af, Wireless LAN Medium Access Control (MAC) and
Physical Layer (PHY) specifications Amendment 2: TV White Spaces
Operation, Version D1.0, January 2011.
Draft ETSI TR 102 907 Reconfigurable Radio Systems (RRS); Use
cases for Operation in White Space Frequency Bands.
Draft ETSI TR 103 067, Reconfigurable Radio Systems (RRS);
Feasibility Study for RF Performance for CRS systems in TVWS.
Draft ETSI TS 102 946, Reconfigurable Radio Systems (RRS); System
requirements for Operation in UHF TV Band White Spaces.
3GPP TS 32.593, Procedure Flows for Type 1 Interface HeNB to HeNB
Management System (HeMS), v10.0.0.
Federal Communications Commission SECOND MEMORANDUM
OPINION AND ORDER, September 23, 2010, available online at:
http://transition.fcc.gov/Daily_Releases/Daily_Business/2010/db0923/F
CC-10-174A1.pdf.
[9]
[10]
[11]
[12]
[13]
[14]
[15]