THE X FACTOR

A THIRD-PARTY BENCHMARK STUDY OF THE

XCOM-LABS’ XCOMP SOLUTION

February 2024

Prepared by
Signals Research Group

www.signalsresearch.com

Study Conducted for XCOM-Labs

As the sole authors of this study, we stand fully behind the results and analysis that we provide in this paper, which leverage a methodology consis-
tent with our benchmark studies that we have conducted for nearly two decades.
In addition to providing consulting services on wireless-related topics, including performance benchmark studies, Signals Research Group is the
publisher of the Signals Ahead research newsletter (www.signalsresearch.com)
 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Key Highlights
Signals Research Group (SRG) conducted a performance benchmark study of the XCOM-Labs’ XCOMP
solution, which is a coordinated multipoint radio system targeting dense deployments of 5G infra-
structure to deliver significant capacity gains over traditional inbuilding solutions. This paper provides
the results from our third-party study which looked at the performance of an XCOMP capable
network designed to support various use cases the company is targeting with its solution.

XCOMP differs from traditional inbuilding solutions, such as small cells and DAS (Distributed Antenna XCOMP provides extremely high
Systems), in that it provides extremely high downlink and uplink capacity that can be both ubiqui- downlink and uplink capacity
tous and highly concentrated in nature. Further, it achieves these performance gains within a single that can be both ubiquitous and
logical cell comprised of multiple RRU (Remote Radio Units), meaning that inter-cell interference is highly concentrated in nature.
nonexistent and brief interruptions due to cell handovers are a thing of the past. Most impressive,
the bidirectional capacity gains we documented were largely immune to the location and concentra-
tion of the devices in the network. We leveraged up to 204 devices in a single test, including a mix of
smartphones and modules, as part of this benchmark study.

Key highlights from our benchmark testing include the following:

XCOMP delivered very high downlink and uplink capacity due to high reuse of nearly all resource
blocks on each MIMO layer. Average downlink capacity reached over 4.5 Gbps in 100 MHz of TDD
spectrum (64.6 bps/Hz) while in the uplink we documented an average throughput of nearly 770 Mbps
(38.3 bps/Hz) with a spectrum allocation that was biased toward the downlink direction. High reuse
of network resources – up to 16 downlink MIMO layers and up to 12 uplink MIMO layers with nearly
full reuse of all resource blocks (RBs) on each layer – largely explains the observed gains. Even more
MIMO layers are possible in one or both directions depending on the network configuration and
device capabilities.

Results were largely impervious to the location and concentration of the devices. In many of our
tests we incorporated 1 to 96 devices located adjacent to each other on up to 12 carts distributed
across the 8 RRU network – up to 204 devices at a time. In one test we pushed all 12 carts together
so there were 204 devices located directly under a single RRU. There were only modest differences in
the total network capacity between these configurations. It is one thing to obtain high bidirectional
spectral efficiency with distributed devices. Achieving this performance with devices literally on top
of each other is another thing altogether.

As an example, in one challenging test with 48 devices grouped together on 6 adjacent carts, the
spectral efficiency during a bidirectional test was an impressive 38.1 bps/Hz (downlink) and 34.5 bps/
Hz (uplink). This high performance was achieved by a near-perfect reuse of all possible RBs in the
downlink and uplink directions along with the complete use of all MIMO layers. In the subsequent
sections, we identify other pertinent information for these tests, such as the channel bandwidth,
downlink/uplink slot allocations, etc. Even with 204 UEs placed under a single RRU, the uplink spectral
efficiency during a test involving uplink performance was an impressive 23.7 bps/Hz with near-perfect
reuse of RBs.

XCOMP is ideal for public venues and private networks that have challenging RF environments.
Stadiums, arenas, and similar venues are ideal for XCOMP since there are high concentrations of
smartphones generating significant data traffic in both directions – fans watching another sporting
event on their phones or concert goers streaming the live concert to all their friends who couldn’t get
a ticket to the show. However, the merits of XCOMP are even more important for private inbuilding
networks targeting factories, warehouses, and other large building structures where there could be a
limited amount of spectrum available and where there is a need for a robust network that provides

February 2024 Page 2

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

highly-reliable ubiquitous coverage while satisfying the demands for concentrated capacity that can
vary over time and location.

Modern warehouses and micro fulfillment centers frequently leverage automation and bots to move
boxes and other containers throughout the inbuilding structure. These buildings can have aisles sepa-
rated by tall shelves which reach the ceiling, and intermixed with small self-contained rooms, such as
refrigeration units, which make it difficult, if not impossible, to deploy seamless connectivity with the
high reliability KPIs which are necessary to keep the automated functions up and running. Further,
with autonomous bots roaming the floors the demands on the uplink performance can take prece-
dence over the downlink requirements. The downlink/uplink capacity requirements of the network
can vary in both time and space, meaning that evenly distributed capacity won’t support high concen-
trations of automated bots while it would be cost prohibitive to deploy sufficient capacity via small
cells, even if the inherent issues due to inter-cell interference could be solved.

Extended Reality (XR) is another use case that places high demands on data capacity. Ideally, this
capacity is not restricted to a single location (e.g., a sofa) since this limitation would prevent more
compelling use cases involving mobility and larger geographic areas. Examples that come to mind
include military and public safety training, not to mention highly realistic gaming and related enter-
tainment activities, such as next-generation laser tag or XR-based haunted houses.

The following sections of this paper provide the results from our benchmark study which we
conducted at the XCOM-Labs’ facility in California. We start off with a short technical description of
XCOMP and how it compares and contrasts with DAS and small cells. The bulk of this paper follows in
the next section which provides the results and analysis of the tests that we conducted as part of this
study. We then include some background information about Signals Research Group.

February 2024 Page 3

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Table of Contents
Key Highlights …………………………………………………………………………………………………………… 2
XCOMP Technology Primer ………………………………………………………………………………………… 6
XCOMP delivers superior downlink and uplink capacity with bidirectional data traffic ………… 8
XCOMP performance is robust and able to support large numbers of devices
clustered together ……………………………………………………………………………………………………… 16
XCOMP achieved these results while also supporting prioritized traffic flows
involving GBR ………………………………………………………………………………………………………… 26
Background …………………………………………………………………………………………………………… 30

Index of Figures
Figure 1. DAS, Small Cells and XCOMP ……………………………………………………………………………… 6
Figure 2. UE Placement …………………………………………………………………………………………………… 8
Figure 3. Network Diagram ……………………………………………………………………………………………… 9
Figure 4. Bidirectional Throughput with 8 Distributed UEs –
SU-MIMO and XCOMP Configurations ……………………………………………………………………………10
Figure 5. Bidirectional Capacity and Spectral Efficiency with 8 Distributed UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 11
Figure 6. Downlink and Uplink Resource Block Allocations with 8 Distributed UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 12
Figure 7. Average Downlink and Uplink Resource Block Allocations with Distributed UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 13
Figure 8. Downlink and Uplink RB Normalized MIMO Layers with 8 Distributed UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 14
Figure 9. Average Downlink and Uplink RB Normalized MIMO Layers with Distributed UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 14
Figure 10. Average MCS Values with 8 Distributed UEs – SU-MIMO and XCOMP Configurations …… 15

Figure 11. Throughput with Distributed and Clustered UEs – SU-MIMO and XCOMP Configurations… 16
Figure 12. Capacity and Spectral Efficiency with Distributed and Clustered UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 17
Figure 13. Resource Block Allocations with Distributed and Clustered UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 18
Figure 14. Average Resource Block Allocations with 8 Distributed UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 18
Figure 15. RB Normalized MIMO Layers with Distributed and Clustered UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 19
Figure 16. Average RB Normalized MIMO Layers with Distributed and Clustered UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 19
Figure 17. Uplink Throughput with 204 Collocated UEs – SU-MIMO and XCOMP Configurations … 20

February 2024 Page 4

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Figure 18. Uplink Capacity and Spectral Efficiency with 204 Collocated UEs –
SU-MIMO and XCOMP Configurations ………………………………………………………………………… 20
Figure 19. Uplink Resource Block Allocations with 204 Collocated UEs –
SU-MIMO and XCOMP Configurations …………………………………………………………………………… 21
Figure 20. RB Normalized MIMO Layers with 204 Collocated UEs –
SU-MIMO and XCOMP Configurations ………………………………………………………………………… 22
Figure 21. Average MCS Values with 204 Collocated UEs – SU-MIMO and XCOMP Configurations 23
Figure 22. Downlink Throughput with 96 Distributed UEs – XCOMP ……………………………………… 24
Figure 23. Downlink Capacity and Spectral Efficiency with 96 Distributed UEs – XCOMP …………… 24
Figure 24. Downlink Resource Block Allocations with 96 Distributed UEs –XCOMP …………………… 25
Figure 25. RB Normalized MIMO Layers with 96 Distributed UEs – XCOMP …………………………… 25
Figure 26. Total Uplink Throughput with 96 UEs – 8 UEs with GBR and 88 UEs with Best Effort …… 26
Figure 27. Average Uplink Throughput with 96 UEs –
8 UEs with GBR and 88 UEs with Best Effort …………………………………………………………………… 27
Figure 28. Total Downlink and Uplink RBs with 96 UEs –
8 UEs with GBR and 88 UEs with Best Effort …………………………………………………………………… 28
Figure 29. Total Downlink and Uplink RB Normalized MIMO Layers with 96 UEs –
8 UEs with GBR and 88 UEs with Best Effort …………………………………………………………………… 29

February 2024 Page 5

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

XCOMP Technology Primer

SU-MIMO (Single-User MIMO) and MU-MIMO (Multi-User MIMO) are similar in that with certain radio
conditions they can reuse the same resource in the time and frequency domain, resulting in higher
data speeds and sector throughput. With SU-MIMO, the network scheduler can simultaneously assign
the same network resource, or Resource Block (RB), to serve a single mobile device. For example, 2x2
SU-MIMO can reuse the same RB twice to nearly double the data speed and 4x4 SU-MIMO can reuse
the same RB four times to theoretically quadruple the data speed.

MU-MIMO is conceptually like SU-MIMO in that it can reuse network resources when certain channel
conditions are satisfied. It differs in that the total number of layers is higher than what is possible with
SU-MIMO and the layers can be shared between multiple mobile devices, assuming they meet certain
algorithmic parameters.

XCOMP is neither DAS nor a small cell architecture, but it leverages some attributes from both solu-
tions, while forgoing their limitations. In fact, it takes the strengths of both solutions and magnifies
their potential in ways that are not possible without the system-level approach that XCOMP invokes.
Figure 1 shows the XCOMP architecture along with DAS and small cells. DAS uses a single source of
network capacity that is then evenly shared across a number of antennas. The coverage is as good as
the layout of the antennas but since the capacity is equally shared across all antennas, it can’t adjust
its capacity to support network traffic that is concentrated in one part of the network. DAS works
great at an airport gate until there is a plane getting ready to board. DAS also doesn’t support capacity
enhancement features, such as MU-MIMO, so it can’t support large amounts of data traffic unless
there is an ample amount of spectrum available.

Figure 1. DAS, Small Cells and XCOMP

Expand capacity
Distributed Antenna System Small Cells XCOMPRUs
XCOMP by just adding
RUs

Core Core gNB

Core
D
gNB U
+
C
U

Source: XCOM-Labs

Each small cell comes with its own source of additional bandwidth, but the bandwidth doesn’t scale
with the increase in the number of small cells. In other words, a 2x increase in the number of small
cells does not double the network capacity since adjacent small cells generate interference with each
other, thus reducing the inherent amount of available capacity. Additionally, cell handovers occur
whenever moving between small cells and these handovers can degrade overall performance and
reliability – especially with a dense concentration of small cells. Isolating the small cells can minimize
inter-cell interference but it comes at the expense of ubiquitous coverage.

XCOMP has a single source of capacity – a distributed unit or DU that is based on the Open RAN
architecture – which combined with a centralized unit (CU) and remote radio units (RRUs) provides
a complete radio access network. The intelligence resides in the DU, meaning that third-party RRUs

February 2024 Page 6

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

can be used in the network if they support certain functionality. XCOMP is really a commercial imple-
mentation of CoMP (Coordinated Multipoint), which is an advanced flavor of MIMO that has been
discussed since the days of LTE. CoMP is also one of the key technology enablers for 5G as it can
extend the capabilities of MU-MIMO, but it also requires a high level of coordination and the avail-
ability of channel state information, not to mention the ability to act on that information.

With XCOMP, all network resources, or resource blocks (RBs), targeting the UEs in the network, are With XCOMP, all resource blocks
transmitted/received through all RRUs in the network. However, since the RRUs are spatially separated (RBs) targeting a particular UE are
and interspersed throughout the network along with the UEs, it is possible to intelligently reuse the transmitted/received through
RBs and their corresponding MIMO layers multiple times. This capability places a very high demand on all RRUs in the network.
network synchronization and transmit power levels to ensure RBs transmitted from each RRU arrive at
their respective UEs at the same time and with appropriate power levels so that they can be decoded
by the UE. Much of the secret sauce behind XCOMP lies in these capabilities and the necessary
messaging between the DU and RRUs that is required to make it all work seamlessly. Any commercial
UE works in an XCOMP network since the UE is unaware of what is occurring within the network.

Since UEs are being served from multiple surrounding RRUs, it is “much easier” to reuse RBs and
increase the number of MIMO layers while maintaining the orthogonality or uniqueness of each trans-
mitted RB. Traditional massive MIMO is similar to XCOMP/CoMP with the biggest difference being
that with massive MIMO there is only a single RRU (cell site) involved in the transmissions. This limita-
tion makes it far more difficult to have good MU-MIMO performance when UEs are located close to
each other. A traditional massive MIMO system that is not based on XCOMP is most appropriate for
outdoor deployments where it becomes more challenging to deploy an XCOMP solution.

One last critical distinction for XCOMP is that it is a single cell network. For example, the 8 RRU network XCOMP is a single cell network.
that we tested as part of this study was comprised of a single PCI (Physical Cell ID). This feature meant
that we never encountered any cell handovers when moving UEs around the network. Cell handover
boundaries are prone to interference from adjacent cells while moving between different cells gener-
ates handovers and their corresponding signaling requirements. In a dense small cell network, the
combination of frequent handovers and large handover zones relative to the coverage area of each
small cell results in degraded performance.

February 2024 Page 7

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

XCOMP delivers superior downlink and uplink capacity with

bidirectional data traffic
In most of our tests, we leveraged RRUs deployed as part of the XCOMP-enabled network to test the
network’s performance with XCOMP functionality disabled as well as enabled. We then compared the
results between the two configurations to demonstrate the performance gains due to XCOMP. With
XCOMP functionality disabled, the network architecture was comparable to traditional DAS (Distrib-
uted Antenna System). It wasn’t logistically feasible to test directly against a small cell architecture,
but we do offer good reasoning for why the XCOMP-enabled network’s performance will be much
better than a small cell architecture in the use cases we tested.

Figure 2 shows the positioning of carts during one of the tests we conducted as well as the very dense
clustering of UEs (modules and smartphones) on the carts. Depending on the test, we used anywhere
from 1 to 12 carts that were also repositioned for various tests as described later in this paper.

Figure 2. UE Placement

RRU

2 0 4 UE s

Source: Signals Research Group

February 2024 Page 8

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Each cart held 12 Android smartphones or 8 modules. Depending on the test, we used anywhere
from 1 UE per cart (module or phone) to all 204 UEs on all 12 carts. We also repositioned the carts
throughout the lab facilities for some of our tests. Figure 3 shows the placement of the RRUs. The
location of the carts with the UEs varied, based upon the test scenario.

Figure 3. Network Diagram

RRU1 RRU2 BBU

RRU3

RRU4

RRU5 RRU8

RRU7

RRU6

Source: XCOM - Labs (recreated by SRG)

February 2024 Page 9

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

As an initial test, we had a high bit rate bidirectional data stream transferring data to and from one
UE on each cart – eight UEs in total. We conducted the test with XCOMP disabled (SU-MIMO) and
with XCOMP enabled. For this test the Band n78 channel bandwidth was 100 MHz with a 7/2/1 slot
allocation, meaning 70% of the bandwidth was allocated to the downlink direction, 20% to the uplink
direction, and 10% for the special slot/guard band.

Figure 4 shows the downlink and uplink throughput during the tests with the two network
configurations. The left figure shows the total throughput, and the right figure shows the average
throughput for each UE in the network. The figure includes acknowledged (ACK) throughput, or
the confirmed throughput at the receive end of the transmission, as well as scheduled throughput,
or the origination throughput at either the UE (uplink) or cell site (downlink).

Figure 4. Bidirectional Throughput with 8 Distributed UEs – SU-MIMO and XCOMP Configurations
Total Cell Throughput Average per UE Throughput

SU-MIMO XCOMP SU-MIMO XCOMP

Configuration Configuration Configuration Configuration
Mbps (100 MHz TDD) (100 MHz TDD) Mbps (100 MHz TDD) (100 MHz TDD)
6000 700
Total Downlink Scheduled Average Downlink Scheduled
Throughput - 8 UEs (Mbps) Throughput - 8 UEs (Mbps)
5000 600

500
4000
Average Downlink ACK
Total Downlink ACK 400 Throughput - 8 UEs(Mbps)
Throughput - 8 UEs (Mbps)
3000
300
2000 Average Uplink Scheduled
200 Throughput - 6 UEs (Mbps)
Total Uplink Scheduled
1000 Throughput - 6 UEs (Mbps)
100
Average Uplink ACK
Total Uplink ACK Throughput - 6 UEs(Mbps)
0 Throughput - 6 UEs (Mbps) 0
0 50 100 150 200 0 50 100 150 200
Time Time
Source: Signals Research Group

4525.0 64.6

6.8x Increase in 38.3

Spectral Efficiency

5.7x Increase in
Spectral Efficiency

663.3 765.6 9.5

6.8
135.5
SU-MIMO XCOMP SU-MIMO XCOMP SU-MIMO XCOMP SU-MIMO XCOMP
Downlink Uplink Downlink Uplink

Total Capacity – Mbps (100 MHz TDD) Spectral Efficiency – bps/Hz (100 MHz TDD)

February 2024 Page 10

 SU-MIMO XCOMP SU-MIMO XCOMP
Configuration Configuration Configuration Configuration
Mbps (100 MHz TDD) The X Factor
(100 MHz TDD) Mbps (100 MHz TDD) (100 MHz TDD)
6000 A third-party benchmark study of the XCOM-Labs’
700XCOMP solution
Total Downlink Scheduled Average Downlink Scheduled
www.signalsresearch.com Throughput - 8 UEs (Mbps) Throughput - 8 UEs (Mbps)
5000 600

500
4000
Average Downlink ACK
As shown in Figure 5, the downlink capacityTotal
withDownlink
XCOMP ACKenabled was over
400 4.5 Gbps and the uplink The downlink capacity
Throughput with XCOMP
- 8 UEs(Mbps)
Throughput - 8 UEs (Mbps)
capacity
3000 was 765.6 Mbps. For the 100 MHz channel, this capacity equated to a downlink spectral enabled was over 4.5 Gbps (64.6
efficiency of 64.6 bps/Hz (6.8x increase over SU-MIMO) and an uplink 300spectral efficiency of 38.3
bps/Hz) and the uplink capacity
bps/Hz (5.7x increase over SU-MIMO). As shown in a subsequent figure, the devices were limited to
2000 Average Uplink Scheduled
200
was 765.6 Mbps (38.3
Throughput - 6 bps/Hz).
UEs (Mbps)
MIMO Rank 2 in the downlink. With MIMO TotalRank
Uplink4Scheduled
enabled on the devices, the downlink capacity
with Throughput
1000 SU-MIMO could have been up to twice - 6 UEs
as high as (Mbps)
shown in the figure. However, with many of
100
the more demanding tests that we conducted it isn’t
Total Uplink ACKclear if increasing the total number of possible Average Uplink ACK
Throughput - 6 UEs(Mbps)
MIMO
0 layers would have benefited overall Throughput
performance. - 6 UEs (Mbps) 0
0 50 100 150 200 0 50 100 150 200
Figure 5. Bidirectional Capacity and Spectral
TimeEfficiency with 8 Distributed UEs – SU-MIMO and XCOMP Configurations Time

Total Cell Capacity - Downlink and Uplink Downlink and Uplink Spectral Efficiency

4525.0 64.6

6.8x Increase in 38.3

Spectral Efficiency

5.7x Increase in
Spectral Efficiency

663.3 765.6 9
9.5
6.8
135.5
SU-MIMO XCOMP SU-MIMO XCOMP SU-MIMO XCOMP SU-MIMO XCOMP
Downlink Uplink Downlink Uplink

Total Capacity – Mbps (100 MHz TDD) Spectral Efficiency – bps/Hz (100 MHz TDD)
Source: Signals Research Group

February 2024 Page 11

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

The next group of figures explains the significant gains in bidirectional throughput due to XCOMP.
First, XCOMP allowed the network to reuse network resources (Resource Blocks) in both directions.
Since each RB carries a data payload, reusing an RB to simultaneously schedule other UEs inherently
increases the total throughput of the network without consuming any additional network resources.
Figure 6 shows the downlink and uplink RB allocations during the test with the two network configura-
tions and Figure 7 provides the average values for the two tests.

When calculating the RBs shown in these figures and in subsequent figures throughout this paper, we
took into consideration the total number of RBs, as well as the number of slots reserved for downlink
and uplink data traffic. In effect, if the network allocated all possible RBs then with a 7/2/1 slot ratio,
70% of the total RBs are PDSCH RBs and 20% of the total RBs are PUSCH RBs.

Figure 6. Downlink and Uplink Resource Block Allocations with 8 Distributed UEs – SU-MIMO and XCOMP Configurations
Total RB Allocations Average per UE RB Allocations

SU-MIMO XCOMP SU-MIMO XCOMP

Configuration Configuration Configuration Configuration
RBs RBs
1600 200

Total DL RBs - 8 UEs Average DL RBs - 8 UEs

1200 150

800 100

400 50
Average UL RBs - 6 UEs
Total UL RBs - 6 UEs

0 0
0 50 100 150 200 0 50 100 150 200
Time Time
Source: Signals Research Group

1488.2

8x Reuse of
PDSCH RBs

6x Reuse of 326.8
185.7 PUSCH RBs
54.6

SU-MIMO XCOMP SU-MIMO XCOMP

Downlink Uplink

Total RBs (100 MHz TDD)

February 2024 Page 12

 The X Factor
800 100
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com
400 50
Average UL RBs - 6 UEs
Total UL RBs - 6 UEs
XCOMP increased the RB reuse by 8x in the downlink direction and by 6x in the uplink. The XCOMP- XCOMP increased the RB reuse by
0 0
enabled DU schedules up to 8 UEs in a single slot, hence the 8x increase in PDSCH RBs suggests a 8x in the downlink direction and
0 reuse of the50downlink RBs.100As configured,
perfect 150XCOMP supported
200 0 uplink MIMO
up to 12 50layers, and 100 150 200
by 6x in the uplink direction.
Time
since the UEs were configured to support MIMO Rank 2, the reuse of PUSCH RBs was limited by the Time

number of concurrent MIMO layers, hence the uplink reuse of RBs was “only” 6x.

Figure 7. Average Downlink and Uplink Resource Block Allocations with Distributed UEs – SU-MIMO and XCOMP Configurations
1488.2

8x Reuse of
PDSCH RBs

6x Reuse of 326.8
185.7 PUSCH RBs
54.6

SU-MIMO XCOMP SU-MIMO XCOMP

Downlink Uplink

Total RBs (100 MHz TDD)

Source: Signals Research Group

Another factor that impacts capacity is the number of MIMO layers. With this test configuration and
8 devices, XCOMP was capable of delivering up to 16 MIMO layers in the downlink and up to 12 MIMO
layers in the uplink direction. As shown in Figure 8 and Figure 9, the results were close to perfect,
with an average of 15.7 RB normalized MIMO layers in the downlink direction and an average of 12 RB
normalized MIMO layers in the uplink direction.

February 2024 Page 13

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

We use the term “RB Normalized MIMO Layers” throughout this paper as it is a term that we always
use in our benchmark studies. When calculating the total number of MIMO layers for each device, we
adjusted its reported number of MIMO layers by the total number of possible RBs. For example, if a
device reported 2 MIMO layers but it only used 50% of the possible MIMO layers then its RB Normal-
ized MIMO layer count would be 1 layer (2 layers x 0.5 = 1 layer). Put another way, achieving 16 MIMO
layers means all possible RBs were being scheduled all the time with each RB using 2 MIMO layers.

Figure 8. Downlink and Uplink RB Normalized MIMO Layers with 8 Distributed UEs – SU-MIMO and XCOMP Configurations
Total RB Normalized MIMO Layers Average per UE RB Normalized MIMO Layers

SU-MIMO Configuration XCOMP Configuration SU-MIMO Configuration XCOMP Configuration

SU-MIMO
RB Normalized Configuration
MIMO Layers XCOMP Configuration SU-MIMO
RB Normalized Configuration
MIMO Layers XCOMP Configuration
16 Normalized MIMO Layers
RB RB
4 Normalized MIMO Layers
16 Total DL RB 4
Normalized
Total DL RB
MIMO Layers - 8 UEs
Normalized
12 3
MIMO Layers - 8 UEs
12 3

8 2
8 2
Average DL RB
Normalized
Average DL RB
4 1 MIMO Layers - 8 UEs
Normalized
4 Total UL RB Normalized 1 MIMO Layers - 8 UEs
Average UL RB Normalized
MIMO Layers - 6 UEs MIMO
Total UL RB Normalized Average ULLayers - 8 UEs
RB Normalized
MIMO Layers - 6 UEs MIMO Layers - 8 UEs
0 0
0 0 50 100 150 200 0 0 50 100 150 200
0 50 100 Time 150 200 0 50 100
Time 150 200
Time Time Source: Signals Research Group

Figure 9. Average Downlink and Uplink RB Normalized MIMO Layers with Distributed UEs – SU-MIMO and XCOMP Configurations
15.9
15.9

12.0
12.0
8x Increase
in RB
8x Increase 6x Increase
Normalized
in RB in RB
6x Increase
MIMO
Normalized Normalized
in RB
Layers
MIMO MIMO
Normalized
Layers Layers
MIMO
Layers
2.0 2.0
2.0 2.0

SU-MIMO XCOMP SU-MIMO XCOMP

SU-MIMO Downlink XCOMP SU-MIMO Uplink XCOMP
Downlink Uplink
Total RB Normalized MIMO Layers (100 MHz TDD)
Total RB Normalized MIMO Layers (100 MHz TDD)
Source: Signals Research Group

February 2024 Page 14

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

The last variable which influences throughput and spectral efficiency is the MCS, or Modulation and
Coding Scheme. Readers that are familiar with MCS know that an MCS value equates to a certain data
payload that can be determined from tables published in 3GPP 5G specifications, with higher MCS
values equating to higher payloads. Figure 10 shows the average downlink and uplink MCS values with
the two network configurations. With XCOMP enabled, the MCS was lower in the downlink while it
was only modestly lower in uplink. Limiting the MIMO layers could have resulted in higher MCS values,
but it could have also come at the expense of overall lower throughput.

Figure 10. Average MCS Values with 8 Distributed UEs – SU-MIMO and XCOMP Configurations

MCS SU-MIMO Configuration XCOMP Configuration

30

25
Average Downlink MCS
20 Average Uplink MCS

15

10

0
0 20 40 60 80 100 120 140 160 180 200
Time

24.3 25.0 25.0

19.7

SU-MIMO XCOMP SU-MIMO XCOMP

Downlink Uplink

Average MCS (100 MHz TDD)

Source: Signals Research Group

February 2024 Page 15

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

XCOMP performance is robust and able to support large

numbers of devices clustered together
The features of XCOMP are best applicable for addressing the capacity needs of large numbers of
UEs, especially UEs that are clustered next to each other. The results in this section demonstrate the
benefits of XCOMP for this use case.

For this first comparative test, we analyzed how XCOMP performed with 48 clustered devices. To
achieve this scenario, we pushed 8 carts next to each other, resulting in a total of 48 devices located
next to each other. The location of the 8 collocated carts was somewhat random, but the carts were
spaced between several nearby RRUs. We used a separate SU-MIMO test involving 96 UEs spread
across the network on 8 carts for comparison purposes, although we note higher throughput might
have been possible with SU-MIMO if the network had been configured to support 4 MIMO layers in
the downlink and 2 MIMO layers in the uplink.

Figure 11 provides a time series plot of the bidirectional throughput with the two network configura- These results are based on
tions and device placements. Figure 12 shows the average throughput and spectral efficiency. For this using a 40 MHz TDD channel
test, and all subsequent tests in this paper, we used a 40 MHz TDD channel in Band n78 with a 5/4/1 with a 5/4/1 slot allocation.
slot allocation. Although the slot allocation still favored the downlink direction, it was more biased to
the uplink direction than commercial 5G networks. However, this slot allocation can be more practical
for industrial applications where there is a greater demand placed on the uplink capacity than the
downlink capacity.

Figure 11. Throughput with Distributed and Clustered UEs – SU-MIMO and XCOMP Configurations
Total Cell Throughput Average per UE Throughput (40 MHz TDD 5/4/1 Slot Allocation)

SU-MIMO 96 XCOMP SU-MIMO 96 XCOMP

Distributed UEs 48 Clustered UEs Distributed UEs 48 Clustered UEs
Mbps Mbps
900 Total DL Scheduled Throughput (Mbps) 20

800 Average DL Scheduled Throughput (Mbps)

700
15
Total DL ACK Throughput (Mbps)
600 Total UL Scheduled Throughput (Mbps) Average DL ACK Throughput (Mbps)
Average UL Scheduled Throughput (Mbps)
500 Total UL ACK Throughput (Mbps)
10 Average UL ACK Throughput (Mbps)
400

300
5
200

100

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Time Time
Source: Signals Research Group

761.9 38.1
34.5

552.4
6.0x Increase 10.5x Increase
in Spectral in Spectral
Efficiency Efficiency
February 2024 Page 16
900 Total DL Scheduled Throughput (Mbps) 20

800 Average DL Scheduled Throughput (Mbps)

The X Factor
700 A third-party benchmark study of the XCOM-Labs’ XCOMP solution
15
Total DL ACK Throughput (Mbps)
600 Total UL Scheduled Throughput (Mbps) Average DL ACK Throughput (Mbps)
www.signalsresearch.com
Average UL Scheduled Throughput (Mbps)
500 Total UL ACK Throughput (Mbps)
10 Average UL ACK Throughput (Mbps)
400
As
300 shown in Figure 12, the downlink capacity increased by 6.0x and the uplink capacity increased by With 48 UEs clustered together,
10.5x with XCOMP. It is also important to remind readers that the XCOMP 5 test results also involved XCOMP delivered downlink spectral
200
48 UEs that were placed adjacent to each other, presumably making it more difficult to achieve high efficiency of 38.1 bps/Hz and uplink
RB
100 reuse with more MIMO layers. The SU-MIMO results stem from devices that were distributed spectral efficiency of 34.5 bps/Hz.
throughout
0 the network. The results, however, suggest that concentrating
0 the devices in a single area
still resulted
0
in very
200
high performance.
400 600 800 1000 0 200 400 600 800 1000
Time Time
Figure 12. Capacity and Spectral Efficiency with Distributed and Clustered UEs – SU-MIMO and XCOMP Configurations
Total Capacity Spectral Efficiency

761.9 38.1
34.5

552.4
6.0x Increase 10.5x Increase
in Spectral in Spectral
Efficiency Efficiency

127.9 6.4
52.6 3.3

96 Distributed 48 Clustered 96 Distributed 48 Clustered 96 Distributed 48 Clustered 96 Distributed 48 Clustered

UEs - Rank 2 UEs - Rank 2 UEs - Rank 1 UEs - Rank 2 UEs - Rank 2 UEs - Rank 2 UEs - Rank 1 UEs - Rank 2
Downlink Uplink Downlink Uplink
Total Capacity – Mbps (40 MHz TDD 5/4/1/ Slot Allocation) Spectral Efficiency – bps/Hz (40 MHz TDD 5/4/1/ Slot Allocation)
Source: Signals Research Group

February 2024 Page 17

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Consistent with the earlier test results, the increase in throughput was due to a combination of high
reuse of network resources (RBs) and an increase in the number of RB normalized MIMO layers. Figure
13 (time series) and Figure 14 (averages) show the downlink and uplink RB allocations for the two test
scenarios. With XCOMP, the PDSCH RBs were reused 7.8x times and the total uplink PUSCH RBs were
reused 5.7x times, compared with SU-MIMO.

Figure 13. Resource Block Allocations with Distributed and Clustered UEs – SU-MIMO and XCOMP Configurations

SU-MIMO Configuration XCOMP Configuration

96 Distributed UEs 48 Clustered UEs
PDSCH/PUSCH RBs Configuration
SU-MIMO XCOMP Configuration
500 96 Distributed UEs 48 Clustered UEs
PDSCH/PUSCH RBs
500
400 Total DL RBs

400 Total DL RBs

300
Total UL RBs
300
200
Total UL RBs
200
100

100
0
0 100 200 300 400 500 600 700 800 900 1000
0
Time
0 100 200 300 400 500 600 700 800 900 1000
Source: Signals Research Group
Time

Figure 14. Average Resource Block Allocations with 8 Distributed UEs – SU-MIMO and XCOMP Configurations
395.6

395.6

7.8x Increase
in PDSCH RBs 236.7
7.8x Increase 5.7x Increase
in PDSCH RBs 236.7
in PUSCH RBs
5.7x Increase
in PUSCH RBs

50.9 41.8

50.9 41.8
96 Distributed UEs - 48 Clustered UEs - 96 Distributed UEs - 48 Clustered UEs -
Rank 2 Rank 2 Rank 1 Rank 2
96 Distributed UEs Downlink
- 48 Clustered UEs - 96 Distributed UEs - Uplink 48 Clustered UEs -
Rank 2 Rank 2 Rank 1 Rank 2
PDSCH/PUSCH RBs (40 MHz TDD)
Downlink Uplink
PDSCH/PUSCH RBs (40 MHz TDD)
Source: Signals Research Group

February 2024 Page 18

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

In addition to high reuse of network resource blocks, the RB normalized MIMO layers increased, as With 48 devices adjacent to each
shown in Figure 15 and Figure 16. Of note, despite 48 UEs located next to each other on 8 adjacent other, XCOMP still delivered
carts, XCOMP delivered close to the theoretical maximum number of RB normalized MIMO layers, or close to the theoretical maximum
15.5 layers in the downlink direction and 11.4 layers in the uplink direction. number of RB normalized MIMO
layers in both directions.
Figure 15. RB Normalized MIMO Layers with Distributed and Clustered UEs – SU-MIMO and XCOMP Configurations

SU-MIMO Configuration XCOMP Configuration

96 Distributed
SU-MIMO UEs
Configuration 48 Clustered
XCOMP UEs
Configuration
RB Normalized96MIMO
Distributed
LayersUEs 48 Clustered UEs
16 Normalized MIMO Layers
RB
16
Total DL RB Normalized MIMO Layers
Total DL RB Normalized MIMO Layers
12
12
Total UL RB Normalized MIMO Layers
Total UL RB Normalized MIMO Layers
8
8

4
4

0
0
0 100 200 300 400 500 600 700 800 900 1000
0 100 200 300 400 500
Time 600 700 800 900 1000
Time Source: Signals Research Group

Figure 16. Average RB Normalized MIMO Layers with Distributed and Clustered UEs – SU-MIMO and XCOMP Configurations
15.5
15.5

11.4
11.4

2.0
2.0 1.0
1.0
96 Distributed UEs - 48 Clustered UEs - 96 Distributed UEs - 48 Clustered UEs -
Rank 2 UEs -
96 Distributed Rank 2 UEs -
48 Clustered Rank 1 UEs -
96 Distributed Rank 2 UEs -
48 Clustered
Rank 2 Downlink Rank 2 Rank 1 Uplink Rank 2
Downlink Uplink
RB Normalized MIMO Layers
RB Normalized MIMO Layers
Source: Signals Research Group

February 2024 Page 19

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Uplink Test with 204 UEs

We next placed 204 devices, including a mix of modules and smartphones, directly under a single
RRU to evaluate how XCOMP performed under an extreme use case. We did an uplink test with this
configuration since the uplink direction is frequently more important in networks supporting factories
and warehouses.

Compared with SU-MIMO, the uplink throughput increased by 7.3x with XCOMP to deliver an uplink
spectral efficiency of 23.7 bps/Hz, even more impressive considering the 204 devices were confined to
12 adjacent carts. Figure 17 and Figure 18 show the uplink throughput from this series of tests.

Figure 17. Uplink Throughput with 204 Collocated UEs – SU-MIMO and XCOMP Configurations
Total Cell Uplink Throughput Average per UE Throughput
SU-MIMO XCOMP SU-MIMO XCOMP
Configuration Configuration Configuration Configuration
Mbps Mbps
450 XCOMP 2.0 XCOMP
SU-MIMO SU-MIMO
Configuration Configuration Configuration Configuration
400
Mbps Total Scheduled Throughput - 204 UEs (Mbps) Mbps
450 2.0
350 1.5 Average ACK Throughput -
400 Total Scheduled Throughput - 204 UEs (Mbps) 204 UEs (Mbps)
300 Total ACK Throughput -
350 204 UEs (Mbps)
250 1.5 Average ACK Throughput -
300 Total ACK Throughput - 1.0 204 UEs (Mbps)
200 204 UEs (Mbps)
250 Average Scheduled
150 1.0 Throughput -
200 0.5 204 UEs (Mbps)
100 Average Scheduled
150 Throughput -
50 204 UEs (Mbps)
0.5
100
0 0
50 0 100 200 300 400 500
0 100 200 300 400 500
0 Time 0 Time
Source: Signals Research Group
0 100 200 300 400 500 0 100 200 300 400 500
TimeEfficiency with 204 Collocated UEs – SU-MIMO and XCOMP Configurations
Figure 18. Uplink Capacity and Spectral Time

Total Uplink Capacity 378.8 Uplink Spectral Efficiency 23.7

378.8 23.7
7.3x
Increase in
Spectral
Efficiency
7.3x
Increase in
Spectral
Efficiency

51.9 3.2

51.9 3.2
SU-MIMO - 204 UEs XCOMP - 204 UEs SU-MIMO - 204 UEs XCOMP - 204 UEs

Total Capacity - Mbps (40 MHz TDD) Spectral Efficiency - bps/Hz (40 MHz TDD)
SU-MIMO - 204 UEs XCOMP - 204 UEs SU-MIMO - 204 UEs XCOMP - 204 UEs

Total Capacity - Mbps (40 MHz TDD) Spectral Efficiency - bps/Hz (40 MHz TDD)
Source: Signals Research Group

February 2024 Page 20

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Uplink Test with 204 UEs

As one might expect, the higher uplink throughput was due to a combination of high RB reuse and
more uplink MIMO layers. In this test, the RB reuse was 7.9x higher than possible with SU-MIMO and
the RB normalized MIMO layer count was 7.9 MIMO layers. In this test the devices were limited to a
single MIMO layer in the uplink direction since the smartphones could only support a single uplink
layer. Since the XCOMP DU schedules 8 UEs in a single slot and each UE was only supporting a single
uplink MIMO layer, the total uplink MIMO layer count was configured for 8 layers. Figure 19 shows the
uplink RB allocations for the two network configurations and Figure 20 provides the total number of
uplink MIMO layers.

Figure 19. Uplink Resource Block Allocations with 204 Collocated UEs – SU-MIMO and XCOMP Configurations

SU-MIMO Configuration
(40 MHz TDD 5/4/1 XCOMP Configuration
Slot Allocation) (40 MHz TDD 5/4/1 Slot Allocation)
PUSCH RBs
400

300

200

100

0
0 50 100 150 200 250 300 350 400 450 500 550
Time

328.9

7.3x Increase
in RB Reuse

41.8

SU-MIMO - 204 UEs XCOMP - 204 UEs

Source: Signals Research Group
Total RBs (40 MHz TDD)

February 2024 Page 21

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Figure 20. RB Normalized MIMO Layers with 204 Collocated UEs – SU-MIMO and XCOMP Configurations

SU-MIMO Configuration XCOMP Configuration

RB Normalized MIMO Layers
10

0
0 50 100 150 200 250 300 350 400 450 500 550
Time

7.9

7.9x Increase in
RB Normalized
MIMO Layers

1.0

SU-MIMO - 204 UEs XCOMP - 204 UEs

Total RB Normalized MIMO Layers (40 MHz TDD)

Source: Signals Research Group

February 2024 Page 22

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

With both network configurations the uplink MCS values remained unchanged, despite the nearly
perfect reuse of uplink RBs and available MIMO layers. The average MCS with SU-MIMO and XCOMP
was 24.9, as shown in Figure 21.

Figure 21. Average MCS Values with 204 Collocated UEs – SU-MIMO and XCOMP Configurations

SU-MIMO Configuration XCOMP Configuration

Uplink MCS
30

25

20

15

10

0
0 50 100 150 200 250 300 350 400 450 500 550 600 650
Time

24.9 24.9

SU-MIMO - 204 UEs XCOMP - 204 UEs

Average Uplink MCS (40 MHz TDD) Source: Signals Research Group

February 2024 Page 23

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Downlink Test with 96 UEs

Although perhaps less interesting given the earlier results, we are including one more set of results in
this section from another test that we conducted as part of this study. In this test, we used 8 modules
on each of the 12 carts, or 96 total UEs. The 12 carts were distributed throughout the network as
shown in Figure 2. The modules were also restricted to a single downlink MIMO layer, which helps
improve overall performance when there is a large number of devices. We once again used a 40 MHz
TDD channel with a 5/4/1/ slot allocation. Readers can also compare these XCOMP results with the
results shown SU-MIMO results provided in Figures 11 through 16.

In this test, the total downlink throughput was 513.6 Mbps, resulting in a spectral efficiency of 25.7
bps/Hz. Figure 22 and Figure 23 support this statement.

Figure 22. Downlink Throughput with 96 Distributed UEs – XCOMP

Total Cell Throughput Average per UE Throughput
Mbps Mbps
600 6
Mbps Mbps
Average Scheduled Throughput – 96 UEs (Mbps)
600 Total Scheduled Throughput - 96 UEs (Mbps) 6
500 5
Average Scheduled Throughput – 96 UEs (Mbps)
Total Scheduled Throughput
Total ACK -Throughput
96 UEs (Mbps)
- 96 UEs (Mbps) Average ACK Throughput – 96 UEs (Mbps)
500 5
400 4
Total ACK Throughput - 96 UEs (Mbps) Average ACK Throughput – 96 UEs (Mbps)
400 4
300 3

300 3
200 2

200 2
100 1

100 1
0 0

0 0 50 100 150 200 250 300 0 0 50 100 150 200 250 300

0 50 100 150 Time 200 250 300 0 50 100 150

Time
200 Source: 250 300
Signals Research Group
Time Time
Figure 23. Downlink Capacity and Spectral Efficiency with 96 Distributed UEs – XCOMP
Total Downlink Capacity 513.6 Downlink Spectral Efficiency
25.7
513.6
25.7

XCOMP (Rank 1) XCOMP (Rank 1)

Spectral Efficiency
XCOMP- bps/Hz
(Rank 1) (40 MHz TDD) XCOMP
Spectral Efficiency (Rank 1) (40 MHz TDD)
- bps/Hz
Spectral Efficiency - bps/Hz (40 MHz TDD) Spectral Efficiency - bps/Hz (40 MHz TDD)
Source: Signals Research Group

February 2024 Page 24

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

During this test, the RB reuse was near perfect, as was the total number of MIMO layers. With the
modules restricted to a single downlink MIMO layer, the best possible outcome was 8 layers. Although
not shown in this paper, the average MCS was 25. Figure 24 provides the information regarding the
PDSCH RBs and Figure 25 shows the MIMO layers during this test.

Figure 24. Downlink Resource Block Allocations with 96 Distributed UEs –XCOMP

PDSCH RBs
700

600

500

400 404.5

300

200

100

0
XCOMP - Rank 1
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320
Time Total RBs (40 MHz TDD)

Source: Signals Research Group

Figure 25. RB Normalized MIMO Layers with 96 Distributed UEs – XCOMP

RB Normalized MIMO Layers

9

8 7.8
7

0
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 XCOMP (Rank 1)

Time Total RB Normalized

MIMO Layers
(40 MHz TDD)
Source: Signals Research Group

February 2024 Page 25

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

XCOMP achieved these results while also supporting

prioritized traffic flows involving GBR
We also did an exploratory test involving 96 devices, of which 8 devices were assigned a guaranteed
bi-directional bit rate (GBR) of 5 Mbps at the application layer while the remaining 88 devices used
best effort. For this test, we used 8 devices on each cart with the distribution of carts documented in
Figure 2. The network (40 MHz TDD, 5/4/1 slot allocation) was configured to support SU-MIMO in the
downlink and XCOMP in the uplink. In order to avoid any confusion and since this benchmark study
was focused on the performance of XCOMP, we are only showing the results for the uplink direction.
We point out the GBR was also maintained in the downlink direction, although given the limitations of
SU-MIMO, the total throughput was relatively low compared to what it would have been with XCOMP.

In this test, we first started the data transfer sessions to the 8 devices with the 5 Mbps GBR (~60 to
~75 seconds) and then switched the data transfer sessions to the other 88 devices. Finally, we had data
transfers occurring with all 96 devices – labeled “Area of Interest” in the following figures. Figure 26
shows the total uplink throughput as well as the combined throughput for both sets of devices. The
figures show that the network successfully provided 5 Mbps in the uplink direction to the 8 phones
with the GBR, while allocating the remaining bandwidth to the 88 other phones on a best effort basis.

Figure 26. Total Uplink Throughput with 96 UEs – 8 UEs with GBR and 88 UEs with Best Effort
Mbps
400
Area of Interest
Total Uplink ACK Throughput -
96 UEs (Mbps)
300
Total Uplink ACK Throughput -
88 UEs (Mbps)
200

100
Total Uplink ACK Throughput -
8 UEs with GBR (Mbps)

0
0 50 100 150 200 250 300 350 400 450
Time

Total: 329.8

Total Uplink ACK Throughput -

88 UEs with GBR (Mbps)
284.6

Total Uplink ACK Throughput -

8 UEs (Mbps) 45.3
Total Capacity - Mbps (40 MHz TDD)
Source: Signals Research Group

February 2024 Page 26

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Figure 27 shows the same basic information, albeit based for the average uplink throughput for each
device in the two groups.

Figure 27. Average Uplink Throughput with 96 UEs – 8 UEs with GBR and 88 UEs with Best Effort

Mbps
7
Area of Interest
6 Average Downlink ACK Throughput - 8 UEs with GBR (Mbps)

4
Average Uplink ACK
3 Throughput -
88 UEs (Mbps)
2

0
0 50 100 150 200 250 300 350 400 450
Time

5.7

3.2

Average Uplink ACK Throughput - 8 UEs with GBR (Mbps) Average Uplink ACK Throughput - 88 UEs (Mbps)

Mbps
Source: Signals Research Group

February 2024 Page 27

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

The network sustained the GBR to the eight devices by allocating those devices more network
resources (RBs), which subsequently increased their RB normalized MIMO layers. Figure 28 shows the
average number of RB allocations on a per UE basis for both sets of phones and Figure 29 provides
the information for the RB normalized MIMO layers. With both parameters, the network biased the
allocation of resources to ensure the GBR was maintained, despite the concurrent data traffic associ-
ated with the additional 88 UEs.

Figure 28. Total Downlink and Uplink RBs with 96 UEs – 8 UEs with GBR and 88 UEs with Best Effort

RBs
350
Total Uplink RBs - 96 UEs Area of Interest
300
Total Uplink RBs -
250 88 UEs

200

150

100

Total Uplink RBs - 8 UEs with GBR

50

0
0 50 100 150 200 250 300 350 400 450
Time

Average Uplink RBs per UE -

88 UEs with GBR
2.8

Average Uplink RBs per UE -

8 UEs with GBR
4.9

RBs
Source: Signals Research Group

February 2024 Page 28

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Figure 29. Total Downlink and Uplink RB Normalized MIMO Layers with 96 UEs – 8 UEs with GBR and 88 UEs with Best Effort

RB Normalized MIMO Layers

9

8 Area of Interest

7 Total Uplink RB Normalized

MIMO Layers - 96 UEs
6
Total Uplink RB Normalized
5 MIMO Layers - 88 UEs
4

2
Total Uplink RB Normalized
1 MIMO Layers - 8 UEs with GBR
0
0 50 100 150 200 250 300 350 400 450
Time

Average Uplink RB Normalized MIMO Layers

per UE - 88 UEs with GBR
0.07

Average Uplink RB Normalized MIMO Layers

per UE - 8 UEs with GBR
0.12

RB Normalized MIMO Layers

Source: Signals Research Group

February 2024 Page 29

 The X Factor
A third-party benchmark study of the XCOM-Labs’ XCOMP solution

www.signalsresearch.com

Background
SRG is a US-based research consultancy that has been in existence since 2004. We publish a subscrip-
tion-based research product called Signals Ahead, which has corporate subscribers that span the
globe and involve all facets of the wireless ecosystem. Our corporate readership includes many of
the largest mobile operators in the world, the leading infrastructure suppliers, subsystem suppliers,
handset manufacturers, content providers, component suppliers, and financial institutions.

One key focus area of our research where we are widely recognized is benchmark studies. These studies
have taken us all over the world to test emerging cellular technologies and features immediately after
they reach commercial status. As an example, since the launch of the world’s first 5G network in 2018,
we’ve published 38 benchmark studies in Signals Ahead pertaining to the next generation technology
through the end of January 2024. These studies have included a wide range of frequencies, device, and
chipset performance, not to mention new features within 5G and how 5G impacts the user experi-
ence with frequently used mobile applications.

Our philosophy in doing benchmark studies is that we are even keeled, data-driven, and as objective
as possible. We present the study’s findings with as much performance data and analysis as possible
and then let the results speak for themselves.

February 2024 Page 30

www.signalsresearch.com