09_02_RA41209EN15GLA1_LTE_Performance_Simulations.pdf

LTE TDD Performance Simulations S imulations

RA41209EN15GLA1

1


RA41209EN15GLA1

2


RA41209EN15GLA1

2


RA41209EN15GLA1

5

LTE TDD Performance Simulations

RA41209EN15GLA1

6


RA41209EN15GLA1

7


RA41209EN15GLA1

8


RA41209EN15GLA1

9


RA41209EN15GLA1

10


RA41209EN15GLA1

11


RA41209EN15GLA1

12


Now let’s have a look on the single user throughput with TxDiversity, 3GPP defines TxDiversity as Transmission mode 2. Tx diversity is the default transmission mode of Packet data shared channel. During the test, the transmission mode was forced to TxDiversity, so it can not automatically changed to MIMO mode.

We have 2 figures for average throughput with Tx diversity, one is for TCP and the other is for UDP. For each scenario, we have unloaded test and 70% loaded test. The purple columns are the result for unload test and the yellow for loaded test. The results shows no much performance deviation between FTP and UDP. Comparing with the theoretical throughput we discussed in previous slides, we can see the throughput in High SINR location can reach 40Mbps, almost half of the theoretical peak value. It should be Ok since 82Mbps theoretical peak value can only be achieved by 2*2 MIMO mode with MCS28. Meanwhile, we can find at the cell edge, for loaded network, the throughput is much lower than that of the unloaded network. It is obviously that the heavy inter-cell interference degrades the cell edge performance much.

RA41209EN15GLA1

13


LTE Spatial Multiplexing may operate in two different modes: open loop spatial multiplexing (open loop MIMO) or closed loop spatial multiplexing (closed loop MIMO). The difference between both modes is that in the case of closed loop MIMO there is more feedback (pre-coding matrix index+CQI+RI) from the UE than in the case of open loop MIMO (CQI/RI). The MIMO Mode Control Functionality, part of RRM, is provided by the eNodeB. It works only in downlink (uplink is not affected) and it switches between single antenna modes, Tx diversity mode and MIMO mode. For the trial test, we forced the transmission mode for OL MIMO for test. The results shows very significant difference on throughput for unloaded and loaded networks, especially at the cell edge. For the loaded network, the throughput at the cell edge is much much lower, even lower than that of the Txdiversity mode. Since we force the TM to OL MIMO, so at the cell edge, although the SINR is bad, but eNodeB still works in MIMO mode with very low modulation and coding scheme, which obviously results in low thoughput. But for the TxDiversity mode, we have 3 dB gain at the cell edge, then much higher modulation and coding scheme can be used, which results in higher throughput. So that’s the reason why we can have better performance at the cell edge with TxDiversity rather than MIMO mode. And also, from the test result, we can see the MIMO mode can only have throughput gain with good SINR, it has low efficiency at the cell edge in loaded networks. It’s consistent with Link level simulations.

RA41209EN15GLA1

14


As it was said before, closed loop MIMO has more feedback (pre-coding matrix index+CQI+RI) from the UE than in the case of open loop MIMO (CQI/RI). According to the test result, we can see for the unloaded network, there is almost no difference on throughput performance. But for the loaded network, the CL MIMO can have advantage on throughput at bad SINR location. We didn’t test the transmission mode 7 (single stream Beamforming mode) since it’s not supported in version RL05TD. It is the feature of RL15TD.

RA41209EN15GLA1

15


For the UL transmission, only single antenna transmission is supported by UE according to LTE specification. And More UL MIMO modes will be realized in LTE-advance later. So for the UL, only the SIMO mode was tested. The graphs below show results for UL SIMO transmission, TCP and UDP traffic respectively, under Loaded (100% UL interference) and Unloaded conditions. At near points, the SINR is very good, the UE can reach the peak throughput about 15Mbps, and very close to the theoretical peak values. But the throughput at the cell edge is extremely low, this is because there are some problems on the slow UL power control algorithms with the software version RL05TD at that time. But anyway we can know the UL interference have great impact on the performance at cell edge. Power control is a mean to improve the cell edge behaviour (improve bit rates and have reliable control plane operation), to reduce inter-cell interference and to reduce the UE power consumption. In the LTE UL, the performance of the system heavily depends on the power control scheme.

RA41209EN15GLA1

16


RA41209EN15GLA1

17


Urban Macro: VoIP

Data

Control

CellRange

UL

DL

UL(64) DL(1024)

FDD

643

680

697

700

TDD

527

680

618

724

707

812

829

TDD(BF)

Coverage PRACH:

PDCCH

864

Max. Data rateMCS Index

PRB #

660m

PDSCH 8x2

661

8.7Mbps

9

100

2x2

668

5.6Mbps

6

100

PUSCH 1x8

638

200kbps

5

6

1x2

636

48kbps

4

2

RA41209EN15GLA1

18


RA41209EN15GLA1

19


RA41209EN15GLA1

20


RA41209EN15GLA1

21


RA41209EN15GLA1

22


RA41209EN15GLA1

23


RA41209EN15GLA1

24


RA41209EN15GLA1

25


RA41209EN15GLA1

26


RA41209EN15GLA1

27


note - MIMO 4x4, theoretical or not implemented (UE implementation is with 2 Rx antennas)

RA41209EN15GLA1

28


RA41209EN15GLA1

29


RA41209EN15GLA1

30


RA41209EN15GLA1

31


RA41209EN15GLA1

32

09_02_RA41209EN15GLA1_LTE_Performance_Simulations.pdf

Recommend Documents