Understanding Beamforming and Massive MIMO

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 1 of 57

SSB Beam sweeping ???
 What is SSB?

— SSB = Synchronization signal (SS)/ PBCH block

— Contains NR-PSS (primary sync channel), NR-SSS (secondary sunc channel), PBCH (broadcast
channel)
— Spans 20 PRBs in frequency and 4 OFDM symbols in time
— Frequency position can be shifted
— SSB burst
— is composed of one or more SSBs
— The maximum number of SSBs within SS burst set, L, for different frequency ranges are
— For frequency range up to 2.4 GHz, L is 4
— For frequency range from 2.4 GHz to 6 GHz, L is 8
— For frequency range from 6 GHz to 52.6 GHz, L is 64
30 kHz (L=8) 0 1 2 3 4 5 6 7

2019-10-18 | | Commercial in Confidence

 SSB sweeping

— Principle — More SSB beams

— Use different beams in different SSB — Advantage:
positions — Higher SINR (higher signal power,
— Steer power in a given direction and reduce lower interference power)
interference in other directions — Drawback
— May degrade initial access latency
— affect UE battery consumption
— Higher overhead
— Conclusion: important to assess tradeoff and
select the right nr of beams

30 kHz (L=8) 0 1 2 3 4 5 6 7

2019-10-18 | | Commercial in Confidence

What can be expected when improving
SSB SNR?
— SSB part of initial access procedure SS/PBCH block
— Improved link budget of initial access procedure NR-PSS, NR-SSS, NR-PBCH
— means more users can see the NR cell
Broadcast system
— Does NOT mean that users will have better information (RMSI, OSI)
throughput or shorter user plane latency
— In Ericsson implementation, data performance is NR-PRACH
decoupled from SSB SNR improvement
→ There is no benefit to overdimension SSB NR-RAR

NR-PUSCH

CRM
(Contention Resolution Message)

2019-10-18 | | Commercial in Confidence

User data beams independent of SSB
beams in E/// software
SSB:
— Up to 8 broad beams must coverage entire cell
— Each beam covering many users
— Not targeting high bit rate per user (10kbps-
100s kbps level)

User data beams (PDSCH and PUSCH):

— Beam targeting each individual user, as
accurate as possible
— Aim is to maximize SINR=>maximize user bit
rate (xMbps-x100sMbps per layer level)

2019-10-18 | | Commercial in Confidence

Overall coverage perspective

Link budget of channels involved in initial access Link budget of data channels

SSB is not the channel with the lowest link budget

→Increasing the number of SSB beams does not increase overall NR coverage

2019-10-18 | | Commercial in Confidence

What channels can be moved to low band
for low- and mid-band CA or DC?
CA DC
SSB — General
MSG1 (PRACH)
MSG2 (PDCCH, PDSCH)
— A main benefit of CA over DC is that one UL suffices ➔
MSG3 (PUSCH) no UL channel is required on mid-band for CA if Pcell is
PDCCH (DCI) on low-band
PUCCH (UCI, SR, HARQ FB)
PDSCH (data) — PDCCH can be moved to low-band for CA if cross-carrier
PUSCH (data) scheduling is supported. Not supported in Rel-15 for
CSI feedback
(PUSCH/PUCCH) bands with different numerology, but will come later
SRS
RLC ack (PUSCH)
SIB1 (PDCCH/PDSCH)

Can be moved
Not in Rel-15
Cannot be moved

2019-10-18 | | Commercial in Confidence

With inter-band DL CA, coverage
limiting channels change
Link budget of channels involved in initial access Link budget of data channels
x dB x dB

— With low+mid band DL carrier aggregation, all channels except DL data channel and corresponding DCI
can be moved to low band
— Only SSB, PDCCH and PDSCH remain in mid band.
→ When combined with inter-band DL CA, SSB sweeping has the potential to improve overall NR coverage
(assuming codebook based DL beamforming, not reciprocity based)
2019-10-18 | | Commercial in Confidence
WHY Beamforming
Electromagnetic
waves and
Information
› Transfers energy and information from one
place to another Claude Shannon (1916-2001)

› The amount of information C [bits/s] that can be

reliably transmitted is a function of the bandwidth
𝑆
𝐶 = 𝐵 ⋅ log2 1+
B and the signal to noise (and interference) ratio 𝑁
S/N

› One main contributor to the noise is thermal

noise: 𝑁 𝑡ℎ𝑒𝑟𝑚𝑎𝑙 = 𝑘𝑇𝐵

𝐶≈𝐵⋅𝑆 when
𝑆
≪1
𝑁 𝑁

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 11

How can w e
i n c r e a s e Bandwidth?
› Quite crowded at 1-3 GHz
300 MHz 3 GHz 30 GHz 300 GHz
› Solution: Use higher frequency bands
– Beyond 6 GHz and up to mmW
› New problem: Higher path loss?
› Solution: Use beamforming

2.1GHz UE2
2.1GHz

Celledge1&2---
Celledge2---

Celledge1---

6,5GHz UE2
UE1 6,5GHz UE1&2 UE1
RBS RBS RBS
< 20% of area

100% of area
Same bit-rate Same bit-rate

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 12

How can w e i m p r o v e
t h e signal quality?
› More power is not really a solution
– we want energy efficient communications

› Better to ensure that the power we use is more efficiently utilized

› Transmit only in directions where the energy will reach the receiver
– Improves the received signal strength
– Reduces interference to other receivers

› Adaptive beamforming and MIMO can be taken much further than in

current systems
– More directive beams
– Multiple beams: serve more simultaneous users, serve each user with more
parallel data streams

› We are still far from utilizing the base station antenna area optimally
C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 13
E v e r y t h i n g is b e a m f o r m i n g
- WELL, aLMosT…

› Sector Antennas
– Static antenna pattern, e.g. 120° azimuth, 10° elevation

› Antenna Tilt
– Very slow changes…

› Tx diversity – not beamforming in LTE

interferer interferer

› Rx Diversity
– MRC, IRC

› Spatial Multiplexing
– Open loop, Closed loop, SU-MIMO, MU-MIMO

› Beamforming, Cell Shaping Tx

– Coverage improvement, energy efficiency… Rx

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 14

D i f f e r e n t levels o f
beamforming

Sectorization 1D Beamforming 2D Beamforming

Massive MIMO

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 15

T h e multi a n t e n n a a d v a n t a g e
Channel capacity (bitrate/Hz) C = log2(1 + SNR)
Capacity proportional to bandwidth
C = BW log2(1 + SNR) (bits/s) Low SNR:

SNR
log(1+SNR) ~ SNR

C
Cap

Beamforming and/or Rx diversity

C


High SNR:
log(1+SNR) ≈ log(SNR)

SNR
C


Power inefficient transmission

C
SNR SNR SNR

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 16

Non-precoded CSI-RS
Combined use of codebook and reciprocity in
NR mMIMO
Type 1 codebook feedback good for SU-MIMO:
• superior coverage and peak rate
Reciprocity good for MU-MIMO Codebook SU-MIMO
• limited range SRS sounding

• With 1Tx4Rx UE (SRS) -> Better peak rate

performance with close loop feedback vs Reciprocity based SU/MU-MIMO

reciprocity.
MU-MIMO with type 2 feedback
Type 2 codebook feedback
• balance coverage and MU-MIMO support
• But UE support not expected before 2020

Flexible and adaptive CSI for DL SU and MU-MIMO over time

Codebook based SU-MIMO with 32-p optimal for initial scenario (coverage and user data rate under low load)
2019-10-18 | | Commercial in Confidence
Type I codebooks
CSI-IM

CSI-RS

— UE estimates the channel based on transmitted

CSI-RS
Codebook
— UE selects rank and precoder weights from the Codebook
Type I codebook
— This is feedback to gNB as RI, PMI and CQI RI, PMI, CQI

— Regular resolution CSI targeting SU-MIMO Codebook = Pre-defined Chose weight to

set of DFT beams maximize | HxW |
— Gives information about the strongest direction
of the channel
— Sufficient for SU-MIMO, but more channel Channel
knowledge typically required for MU-MIMO estimate [H]
(Type II codebooks)

2019-10-18 | | Commercial in Confidence

 No precoded CSI-RS: TypeI codebooks (E///)
– CSI-RS ports are assumed to be organized either on a line or in a 2D grid (3GPP spec)
– N1, N2 ports (per polarization) along the “first”and “second” dimension (O1, O2 are the oversampling factor)
– Physical antennas can be more than virtual ports (CSI-RS ports): e.g. 64 TRX > 32,16,8 csi-rs ports
– P2A mapping: each CSI-RS port is mapped to a group of physical sub-arrays (P2A is vendor implementation)
– PMI feedback -> i_11 and i_12 (direction on first and second dimension), i_13 info on second beam direction

E/// config in green 8-p N1, N2 = 4,1

Number of (N1, N2) (O1, O2) Beams — X

CSI-RS 8-p N1, N2 = 2,2
antenna ports P2A
4 (2,1) (4,1) — X
(2,2) (4,4) (8x8)=64
8 16-p N1, N2 = 8,1
(4,1) (4,1) (16x1)=16
(3,2) (4,4)
12
(6,1) (4,1) — X
(4,2) (4,4) (16x8)=128 16-p N1, N2 = 4,2
16
(8,1) (4,1) (32x1=32
(4,3) (4,4) — X (CSI-RS ports)
24 (6,2) (4,4)
(12,1) (4,1) 32-p N1, N2 = 8,2
(4,4) (4,4) 6488 -> 64TX (32TX per pol.)
32 (8,2) (4,4) (32x8)=256 — X
(16,1) (4,1)
2019-10-18 | | Commercial in Confidence
 No precoded CSI-RS: TypeI PMI codebooks (E///)
– CSI-RS ports are assumed to be organized either on a line or in a 2D grid (3GPP spec)
– N1, N2 ports (per polarization) along the “first”and “second” dimension (O1, O2 are the oversampling factor)
– Physical antennas can be more than virtual ports (CSI-RS ports): e.g. 64 TRX > 32,16,8 csi-rs ports
– P2A mapping: each CSI-RS port is mapped to a group of physical sub-arrays (P2A is vendor implementation)
– PMI feedback -> i_11 and i_12 (direction on first and second dimension), i_13 info on second beam direction

E/// config in green 8-p N1, N2 = 4,1

Number of (N1, N2) (O1, O2) Beams — X

CSI-RS 8-p N1, N2 = 2,2
antenna ports P2A
4 (2,1) (4,1) — X
(2,2) (4,4) (8x8)=64
8 16-p N1, N2 = 8,1
(4,1) (4,1) (16x1)=16
(3,2) (4,4)
12
(6,1) (4,1) — X
(4,2) (4,4) (16x8)=128 16-p N1, N2 = 4,2
16
(8,1) (4,1) (32x1=32
(4,3) (4,4) — X (CSI-RS ports)
24 (6,2) (4,4)
(12,1) (4,1) 32-p N1, N2 = 8,2
(4,4) (4,4) 6488 -> 64TX (32TX per pol.)
32 (8,2) (4,4) (32x8)=256 — X
(16,1) (4,1)
2019-10-18 | | Commercial in Confidence
 No precoded CSI-RS: TypeI PMI codebooks (E///)
– CSI-RS ports are assumed to be organized either on a line or in a 2D grid (3GPP spec)
– N1, N2 ports (per polarization) along the “first”and “second” dimension (O1, O2 are the oversampling factor)
– Physical antennas can be more than virtual ports (CSI-RS ports): e.g. 64 TRX > 32,16,8 csi-rs ports
– P2A mapping: each CSI-RS port is mapped to a group of physical sub-arrays (P2A is vendor implementation)
– PMI feedback -> i_11 and i_12 (direction on first and second dimension), i_13 info on second beam direction

E/// config in green 8-p N1, N2 = 4,1

Number of (N1, N2) (O1, O2) Beams — X

CSI-RS 8-p N1, N2 = 2,2
antenna ports P2A
4 (2,1) (4,1) — X
(2,2) (4,4) (8x8)=64
8 16-p N1, N2 = 8,1
(4,1) (4,1) (16x1)=16
(3,2) (4,4)
12
(6,1) (4,1) — X
(4,2) (4,4) (16x8)=128 16-p N1, N2 = 4,2
16
(8,1) (4,1) (32x1=32
(4,3) (4,4) — X (CSI-RS ports)
24 (6,2) (4,4)
(12,1) (4,1) 32-p N1, N2 = 8,2
(4,4) (4,4) 6488 -> 64TX (32TX per pol.)
32 (8,2) (4,4) (32x8)=256 — X
(16,1) (4,1)
2019-10-18 | | Commercial in Confidence
Azimuth Mapping of Parameter i11 for 64Tx and
32Tx Radios, N1=8

2019-10-18 | | Commercial in Confidence

Elevation Mapping to Parameter i12 for 64Tx
Radios, N2=2

2019-10-18 | | Commercial in Confidence

6488 Traffic envelope: 32 ports (8,2) and 16 ports (8,1)

CSI-RS with 32 and 16 achieves already the full BF capability on azimuth

2019-10-18 | | Commercial in Confidence

Preliminary Results From OTA (Kista): 8 vs 32 csi-rs

Field test results are aligned with cov. simulations

- Similar performance at high SINR
- 32-ports has better thpt at mid to low SINR
- No issue for 32-ports at cell edge

2019-10-18 | | Commercial in Confidence

 PMI feedback – Example 4 UEs MU-MIMO (32-p)
- The PMI feedback contains 4 possible values I11, I12, I13 and I2. Currently I2 is not used. The values
represents the following:

▪ I11 : first layer azimuth, number of values possible depend on panel geometry in dimension 1
▪ I12 : first layer elevation, number of values possible depend on panel geometry in dimension 2
▪ I13 : different pointing for other layers (used only if more than 1 layer)
▪ I2 : polarization co-phasing term (not used at the moment)

UE UE

UE

UE

Tx

Kista OTA
i1,1=0 & i1,2=0 -> boresight
2019-10-18 | | Commercial in Confidence
M -M IMO
SU-MIMO v s
MU-MIMO
 WHAT IS IT?

Single-User MIMO Multi-User MIMO

— One user per time-frequency resource — Multiple users on the same resource
— User specific BF provide array gain — User specific BF to spatially separate users

highly focused beam

— One User is served faster. Hence, the cell capacity is — Many users can be served at the same time thanks to the
increased significantly increased capacity that MU-MIMO provides
— Users in the cell edge will get a much higher throughput
 MASSIVE MIMO
MIMO L a y e r s

Single-User MIMO Multi-User MIMO

— The Scheduler decides how many MIMO Layers that the — As in SU-MIMO, the Scheduler decides how many MIMO
beam will be set up with. Layers that the beam will be set up with.

e.g. SU MIMO 64x2

— Many terminals support up to 2 MIMO Layers only, rank 2 — How many simultaneous users that can be supported is
(limited by the chip-set in the terminal) defied by the number of MIMO Layers that the system
supports
S i n g l e u s e r MIMO (SU-MIMO)

Layer 1
› In SU-MIMO one user per time-
UE 1 frequency resource on all layers
UE 2

Layer 2 UE 3 › User specific Beamforming provide

array gain
UE 4
time
› SINR increases as #antennas
increase
› Benefits regardless of load

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 31

M u l t i u s e r MIMO (MU-MIMO)

› In MU-MIMO multiple users are using the

Layer 1
UE 1 same time-frequency resources

UE 2
› MU-MIMO is beneficial if
Layer 2 UE 3 – UE is BW limited, i.e. has maxed out it’s
capacity (in good SINR)
UE 4 – More layers available than UE capability

Layer 3
› MU-MIMO prerequisite
– there are UEs to “pair” and
– they are spatially separated and
Layer 4 – the combined bit rate to the spatially separated
UEs is higher than then the bit rate a single UE
could get

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 32

M u l t i u s e r MIMO (MU-MIMO)

Layer 1
TM8 (64x2 MIMO)
Rank 2 capable devices

Layer 2

Separated by
beamforming
Layer 3

64T64R

Layer 4
UE specific
beams

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 33

MU-MIMO s c h e d u l i n g E x a m p l e
Scheduling example
Layers

UE3 UE3 UE3

UE2 UE2 UE2 UE3 UE4 UE4

UE1 UE1 UE1 UE1 UE3 UE3 UE4

RBG
UE1 needs 4 RBGs to empty its buffer
UE2 needs 3 RBGs to empty its buffer
UE3 needs 6 RBGs to empty its buffer
UE4 needs 3 RBGs to empty its buffer and does not meet the channel separation criteria with UE1
C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 34
M a s s i v e MIMO: SU-MIMO
a n d MU-MIMO Ex a m p l e
› Full Ericsson Asian scenario
– 42 sectors in central UE dense part
– 63 sectors surrounding
› All 105 sectors use the same
antenna setup
› All 10000 users included
› FTP1 traffic model, 100 kByte
› cell-specific tilt used for each cell

SU-MIMO

Massive MIMO gain is a combination of SU and MU-

MIMO. In this high load scenario: MU-MIMO≤8 layers

• 50% transmissions with SU-MIMO

• 47% transmissions with MU-MIMO ≤8 layers
• MU-MIMO>8 layers has <3%

C | 2020-02-26 | AAS & Massive MIMO BASICS | Ericsson Confidential | Page 28

 MU-MIMO

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 36 of 57

 DL MU-MIMO

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 37 of 57

 UL-MIMO

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 38 of 57

 MU-MIMO Problem
Statement

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 39 of 57

 MU-MIMO Testing Methodology
● Keep 4 UEs at a angular horizontal
spacing with 25 to 30 deg separation
with Good RSRP >-95dBm

● Check for MU-Users are pegging

beyond User 3 and user 4

● Compare the 15 ROP stats with and

with MU-MIMO activated

● FTP download with 10 sessions per

UE

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 40 of 57

 Statistics
%DL %DL %DL %DL %DL %DL %DL %DL %DL %DL %DL %DL
DL DL Layer1 Layer2 Layer3 Layer4 Layer5 Layer6 Layer7 Layer8 User1 User2 User3 User4
Test UE
Volume( PRB DL UE DL CELL DL Avg Sample Sample Sample Sample Sample Sample Sample Sample Sample Sample Sample Sample
GB) Util THP THP MCS CQI s s s s s s s s s s s s
32T 4 98.57 69.40 520.87 1335.70 25.89 13.20 0.80 2.58 32.36 64.26 0.00 0.00 0.00 0.00 100.00 0.00 0.00 0.00
32T- MU-
4
MIMO 88.11 9.97 244.15 1163.04 17.20 11.59 0.14 1.76 3.80 48.33 0.31 45.61 0.00 0.06 12.12 41.91 45.92 0.06
32T 12 102.90 84.04 116.49 1345.58 24.49 12.63 0.11 0.16 29.60 70.12 0.00 0.00 0.00 0.00 100.00 0.00 0.00 0.00
32T- MU-
12
MIMO 88.00 39.12 78.77 1169.84 17.72 13.61 0.56 2.02 2.75 39.05 1.45 44.08 0.17 9.92 8.64 35.64 45.60 10.12
32T 27 64.33 90.15 38.72 821.01 18.78 9.38 0.34 53.90 29.38 16.38 0.00 0.00 0.00 0.00 100.00 0.00 0.00 0.00
32T- MU-
27
MIMO 49.41 57.13 26.92 667.70 15.13 9.22 7.66 3.68 53.65 33.65 0.22 1.13 0.00 0.00 62.61 36.02 1.37 0.00

●Observation with significant % of number layers on 6

1. PRB utilization goes down by more than 50% with Mu_MIMO activated
2. DL volume and cell throughput reduced with MU_MIMO activation
3. SU_MIMO shows better throughput than MU_MIMO

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 41 of 57

 Problem Statement

1. Why Cell throughput with MU-MIMO is lesser than SU-MIMO with higher
number layers scheduled

2. Reason for huge reduction PRB Utilization under MU-MIMO

3. Reason for reduction in MCS under MU-MIMO, When the CQI is same

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 42 of 57

 MU-MIMO Vs SU-
MIMO Gain

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 43 of 57

 Executive Summary

DL – MU-MIMO MU-MIMO gain

cell throughput gain depends on the
achieved upto 2.3x orthogonality and
times UE spread

In the commercial
UL- MU-MIMO cell network the MU-
throughput gain MIMO gain
achived upto 2.3x depends on the UE
gain. spread and its
orthgonality

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 44 of 57

Testing Information -1
● Site – 4858 Sector C
● Azimuth: 10deg
● Number of UE : 4
● Software – 21.Q2
● Date : 4th Dec 2021

● Timing – Rop interval

Start
Antenna Date UE End Time
Time

32T_MU 4-Dec-21 4 3:45PM 4:00PM

32T 4-Dec-21 4 4:30PM 4:45PM

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 45 of 57

 MU-MIMO Gain
Average DL CELL THP In Mbps DT - DL UE Thpt
3000.0 600.0
2632.5 529.2
2500.0 500.0
2.2x Gain 400.0
2.3x Gain
2000.0

1500.0 300.0 232.1

1196.2
1000.0 200.0

100.0
500.0
0.0
0.0
32T_MU 32T
32T_MU 32T

DL Volume(GB)
250.0
202.6
200.0
2.3x Gain
150.0

100.0 87.4

50.0

0.0
32T_MU 32T

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 46 of 57

 MU Layers and Users

MU-MIMO Layers MU-MIMO Users

84.25
90.00 120.00
80.00 100.00

62.99
100.00
70.00
84.25
60.00 80.00
50.00
36.97

40.00 60.00

30.00
40.00

11.71
20.00
2.89
0.66

0.48

10.00 20.00
0.03
0.01

11.71
0.00 1.97 2.07
%DL Layer1 %DL Layer2 %DL Layer3 %DL Layer4 %DL Layer5 %DL Layer6 %DL Layer7 %DL Layer8 0.00
Samples Samples Samples Samples Samples Samples Samples Samples %DL User1 Samples %DL User2 Samples %DL User3 Samples %DL User4 Samples

32T_MU 32T 32T_MU 32T

More than 80% of the time 4 MU-MIMO users are scheduled

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 47 of 57

Testing Information - 2

● Site – 9014 Sector B

● Number of UE : 4
● Software – 21.Q2
● Antenna comparison – 8T vs
32T MU

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 48 of 57

 DL Throughput, Cell Throughput and Volume

1.8x Gain 1.9x Gain

1.9x Gain

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 49 of 57

 Coverage Details

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 50 of 57

MU-Users and Layers

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 51 of 57

 UL – MIMO

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 52 of 57

Testing Information -1
● Site – 4858 Sector C
● Azimuth: 10deg
● Number of UE : 4
● Software – 21.Q2

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 53 of 57

 UL Thpt – UL cell Thpt - UL Volume – 4858

2.4x Gain 2.3x Gain 2.2x Gain

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 54 of 57

 CQI-UL MCS-UL SINR

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 55 of 57

MU - Users and Layers

| EILMAMO Magesh Mohan | 2021-12-01 | Ericsson Confidential | Page 56 of 57