Академический Документы
Профессиональный Документы
Культура Документы
2018-07-03
2018-07-05 | Commercial in confidence | Page 1
Overview
Codeword 1
IQ IQ
Layer 1
codeword
ഥ w11 w12
𝑿 ഥ =𝑾∙𝑿
ഥ
𝒀
𝑤11 𝑤21 Antenna port 0
𝑾= 𝑤
Layer 2
codeword
12 𝑤22 Antenna port 1
w21 w22
Codeword 1 Codeword 2
12 sub-carriers
— For CFI3 TM8 as 10% less REs available for PDSCH
than does TM3
— The TM8 beamformed PDCCH will have a better SINR
which can lead to higher throughputs than TM3/TM2 at
medium to low SINR
RS0 RS1
PDCCH PDSCH PDCCH or PDSCH based on CFI
2018-07-05 | Commercial in confidence | Page 6
R7 & R8 UE-specific (DM-RS)
AAS Cell Coverage
EIRP max per broadcast beam 2 x 64.5 dBm 2 x 71.5 dBm 2 x 64.5 dBm
Digital downtilt
2018-07-05 | Commercial in confidence | Page 8 -8° to +8° n/a n/a
AAS Antenna Beam Examples
AIR6468 Band 42 Horizontal Vertical
Macro Broadcast Beam, Vtilt= 8
Scenario: Macro Beam
Parameter BrM1, BrM2
Vertical Beamwidth 10±1
Horizontal beamwidth 65±5
Digital downtilt [-8,8]
Vertical beam pointing error ≤1
Horizontal beam pointing 0±5
EIRP (max) 2x64.5 dBm
Vertical side lobe suppression 16 dB
Front to back ratio 25 dB
Beam parallellity (BrM1•BrM2) ≤-10 dB
Traffic Beam, Htilt=0,Vtilt= 3
B
— An increased number of MU-MIMO users increases the
interference between users/layers
— Since power is shared by the co-scheduled layers the
power per layer reduces as the number of users MU-
MIMO increases EIRP
— SINR reduces as the number of MU-MIMO co-scheduled
users increases A B
SINR
Frequency reuse
(MU-MIMO)
SINR q
(MCS per MIMO layer)
Interference from B into A
Typically # layers
2018-07-05 | Commercial in confidence | Page 11
~8 layers
Antenna Pattern
Multiple Traffic Beams Horizontal Pattern
14 REs
12 sub-carriers
MCS Modulation TBS Physical Resource Blocks (RB)
Index Index
TBS 1 2 .. 50 .. 100
0 QPSK 0
0 16 32 1384 2792
15 16QAM 14
14 256 552 14112 28336
27 64QAM 25
28 64QAM 26
25 616 1256 31704 63776 2 Tx antenna – CFI = 1 4 Tx antenna – CFI = 1
26 712 1480 36696 75376
1 RB = 144 RE for PDSCH 1 RB = 136 RE for PDSCH
12 sub-carriers
— TM8 introduces DMRS to assist with the UE with demodulation of the
beamformed PDSCH
— The PDSCH and DMRS are transmitted on the traffic beam
— The DMRS reduces the RE available for PDSCH limiting the peak rate RS0 RS1
— The UE does not demodulate a coding rate > 0.93 PDCCH PDSCH PDCCH or PDSCH based on CFI
TBS 25 ≈ [63776+24] / [10800x6] Coding rate ≈ 0.98 63,776 x 2 x 0.74 = 94.4 Mbps
Peak rates use MCS 28
TBS 24 ≈ [61664+24] / [10800x6] Coding rate ≈ 0.95 61,664 x 2 x 0.74 = 91.3 Mbps
Bits / TTI
64QAM Modulation
2018-07-05 | Commercial in confidence& | Page 14 TBS 23 ≈ [57336+24] / [10800x6] Coding rate ≈ 0.88 57,336 x 2 x 0.74 = 84.9 Mbps
MIMO Transmission
Transmission Modes
Scheduling SU-MIMO & MU-MIMO
The following applies when a cell is configured for TM8 transmissions and
transmission mode switching is not activated >15°
— The eNB checks the UE pairs with data in the buffer for beam-weight
orthogonality and the assignment of SRS resources >15°
— If MU-MIMO is not required for “capacity” reasons and the UEs have SRS >15°
resources assigned then the data is scheduled as TM8 Dual Layer
— The eNB “estimates” the beam weights for the other polarization
— If MU-MIMO is required for “capacity” reasons and the UEs have SRS resources
UEs need angular separation in order to be
then the data is scheduled as TM8 Single Layer scheduled for MU-MIMO, otherwise there
— The UEs co-scheduled on the same PRB/TTI resources have different beam weights would be “traffic beam overlap” leading to
applied to the transmissions of the PDSCH and DMRS high levels of intra-cell interference.
— If the UEs don’t have SRS resources assigned then data is transmitted as TM3
SRS configured UEs with angular separation in a UEs in an AAS cell which is not configured
cell requiring MU-MIMO for capacity may be co- for TM8 beamforming will receive DL
scheduled on the same TTI/PRB resources. These transmissions in TM3 MIMO or Tx Diversity.
UEs will equally share the power available for The UEs will share the PRBs &/or TTIs.
PDSCH and DMRS REs and will have orthogonal
beam-weights applied to their transmissions.
AAS
Non-AAS
UEs in a non-AAS cell will receive DL
transmissions in TM3 MIMO or Tx Diversity. A UE in a AAS cell configured for TM8 which does not required
MU-MIMO transmissions for “capacity” reasons will have data
scheduled in TM8 Dual Layer (provided there is sufficient data to
required 2 layers and the SINR is high enough for 2 layer
transmission, otherwise TM8 Single Layer will be used).
2018-07-05 | Commercial in confidence | Page 16
Demonstrating Peak Throughput
Maximizing MU-MIMO Performance
— How to demonstrate the maximum MU-MIMO capabilities for showcase demos
— UEs are separated so that the angular separation between the UEs is ≥ the half power
beamwidth of the traffic beam
— Ideally an single cell scenario is used do UEs wont connect to another sector of the same Front view
site
Side view
>10°
196m 107°
152m 121°
UE6
UE7
166m 117°
UE5
147m 126°
UE3
89m 93°
UE2
85m 113°
UE Number TM3 – MU- TM8 – MU- UE4
MIMO OFF MIMO ON 85m 173°
1 13 45 UE1
2 15 42 47m 150°
3 14 45
4 12 45
5 15 40
6 13 46 AAS
7 13 42
8 15 45
2018-07-05 | Commercial in confidence | Page 19
Total 110 Mbps 351 Mbps
Cell Performance
Monitoring
Throughput - Experience
Reliability of Throughput
measurement to show potential
Throughput for a user
Very small burst like TCP ACKs and 32B Pings etc.
How is throughput measured for a 54B Ping that is
sent in 1 TTI ?
- 432 Kilobits/s?
pmPdcpVolDlDrb = Σ + pmPdcpVolDlDrbLastTTI = Σ
pmUeThpTimeDl = count ( , ) pmSchedActivityCellDl = count ( , , )
Data arrives to
Grant sent to UE Time (ms)
empty UL buffer First data is The send buffer is
Scheduler request transmitted to the eNB again empty
sent to eNB
pmUeThpVolUl = Σ pmPdcpVolUlDrb = Σ + +
pmUeThpTimeUl = count ( , ) pmSchedActivityCellUl = count ( , , , )
— Cell B48752
— Date: Friday 29/06/2018, hourly data
— Busy hour RRC Connected Users ≈ 40 users
— DL Cell throughput > 30 Mbps
— DL UE throughput generally ~ 10 Mbps
— At 20:00 dropped to around 6 Mbps
— Max RRC Connected Users close to average
Counters
Counters
Counters
— The video was 1.04 min in duration, streamed on a commercial mobile 3.5
3.0
network 2.5
2.0
— The Chrome browser was used and the protocol used in the video stream 1.5
1.0
was QUIC (Quick UDP Internet Connections, pronounced quick) which is an 0.5
experimental transport layer network protocol designed by at Google 0.0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
Elapsed Time [Sec]
— Total data volume for 360p
Inbound 4.66 MB, Data Volume vs Time - 1.04 min Youtube 1080p Video
24
Outbound 198.5 kB 22
Ratio DL:UL 23.5
Counters
Counters
Layer1 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[0]
Layer2 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[1]
Layer3 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[2]
Layer4 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[3]
Layer5 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[4]
Layer6 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[5]
Layer7 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[6]
Layer8 = EUtranCellTDD.pmPrbUsedMimoLayersDlDistr[7]
2018-07-05 | Commercial in confidence | Page 31
Cell Deep Dive
DL Transmission Modes
Tm2 = EUtranCellTDD.pmRadioTxRankDistr[0]
Tm3_Rank1 = EUtranCellTDD.pmRadioTxRankDistr[1]
Tm3_Rank2 = EUtranCellTDD.pmRadioTxRankDistr[2]
Tm8_Rank1 = EUtranCellTDD.pmRadioTxRankDistr[6]
Tm8_Rank2 = EUtranCellTDD.pmRadioTxRankDistr[7]
2018-07-05 | Commercial in confidence | Page 32
Mapping CQI to SINR
Values Based on Field Testing from a Network
Typical mappings This is a typical mapping of CQI to SINR based
CQI index SINR [dB] on field test of an early LTE terminal.
0 out of range
1 -7.23 Chipset manufacturers can have different
2 -5.48 mapping as this mapping is not standardized.
3 -3.43
4 -1.37
5 0.55
6 2.47
7 4.58 Example CQI stats for a FDD three
8 6.39 sector site in a MBB network
9 8.34
10 10.44
11 12.39
12 14.15
13 16.02
14 17.61
15 18.95
Cell Deep Dive
UE CQI Reports
Counters
Traditional TM3 FWA cell
DL Bits Per RE = EUtranCellTDD.pmRadioThpVolDl / EUtranCellTDD.pmRadioThpResDl
UL Bits Per RE = EUtranCellTDD.pmRadioThpVolUl / EUtranCellTDD.pmRadioThpResUl
Modulation Format
— The UE CQI reports are used to determine the MCS that the
BTS uses to transmit to the UE 16QAM
— An average of 3 bits per RE indicates 64QAM modulation is
used on average which matches the modulation format QPSK
indicator from the counters
— For TM8 CFI3 there are 10,800 REs in 100 RB CQI index modulation code rate x 1024 efficiency
0 out of range
— So potential TM8 Rank 2 transmission at 3 bits/RE is: 1 QPSK 78 0.1523
2 QPSK 120 0.2344
2 x 10,800 x 3 = 64,800 bits in 1 ms 3 QPSK 193 0.3770
4 QPSK 308 0.6016
5 QPSK 449 0.8770
6 QPSK 602 1.1758
7 16QAM 378 1.4766
Counters 8 16QAM 490 1.9141
9 16QAM 616 2.4063
QPSK = EUtranCellTDD. pmMacHarqDlAckQpsk
10 64QAM 466 2.7305
16 QAM = EUtranCellTDD. pmMacHarqDlAck16qam 11 64QAM 567 3.3223
64 QAM = EUtranCellTDD. pmMacHarqDlAck64qam 12 64QAM 666 3.9023
13 64QAM 772 4.5234
According to 3GPP TS 36.213 a CQI of 10 maps to a code rate of 466 or an efficiency of 14 64QAM 873 5.1152
2.7305, while a CQI of 11 maps to a code rate of 567 or an efficiency of 3.3223. 15 64QAM 948 5.5547
2018-07-05 | Commercial in confidence | Page 37
It should be noted however, that CQI calculation may vary from different UE vendors.
Cell Deep Dive
PDCCH CFI Format
Counters
pmRadioRecInterferencePwrPrb0 … 99
— Wide band measurements – one sample every 1ms placed into the PDF bin ranges shown, a weighted average
is taken using a linear value for the bin ranges
— Sub frame measurements – one sample taken every 1ms and the power level placed in an accumulation counter.
The accumulation value is in pW (1 x 10-9) with 90,000 samples taken in a 15 minute measurement period. Counters are
available for all possible UL subframes.
— PRB measurements – one sample taken every 40ms for each PRB and the power level placed in an accumulation counter for
each PRB. The PRB counters are numbered 0 to 99. The unit of the counter is 1 mW * 2^(-44).
— pmRadioRecInterferencePwrPrb0 .. 99