SA Vol 32 4CC Uplink

Redefining Research May 10, 2023
VOLUME 32: UP
A BENCHMARK STUDY OF 5G mmWAVE
FOUR COMPONENT CARRIER UPLINK PERFORMANCE
PART OF “THE MOTHER OF ALL NETWORK BENCHMARK TESTS” SERIES OF REPORTS
YOUR ATTENTION
PLEASE
EFFECTIVE IMMEDIATELY, ANY UNAUTHORIZED USE OF OUR RESEARCH MATERIAL WILL RESULT IN THE NON-REFUND-
ABLE CANCELLATION OF YOUR SUBSCRIPTION. We also reserve the right to post your company’s name, with logo, to the
“SRG Wall of Shame.” If you received this issue from someone outside of your organization and it did not come directly from
SRG then the licensing terms for our research are being violated. If you forward this research to external organizations, either
in whole or in part, or if you share the contents of the report beyond the authorized allocation within your organization then
the licensing terms for our research are being violated.
If you value the information and insight that we provide then I strongly urge you to respect our hard work and livelihood and
subscribe to our research. If you do not have a platinum license or a global license, you may want to upgrade your license so
that you can share this issue across your entire organization with our blessing.
If you or your organization is interested in distributing this report to outside organizations, please feel free to contact us to
discuss licensing terms and fees.
If you would like to leverage a quote from this report and you have at least a global license, please contact us for permission
and we will be happy to provide it.
1.0 Executive Summary
Key Highlights from this Study

SRG just completed its 32nd 5G benchmark study. For this report, we conducted a bench-
mark study of 4 component carrier (4CC) 5G mmWave uplink performance on AT&T’s
Band n260 (39 GHz) network in Glendale, Arizona, the home of State Farm stadium and
the Westgate Entertainment District. As part of this study, we attended a US-Mexico
soccer match, which gave us full access to the stadium which recently hosted the Super
Bowl and Taylor Swift’s first stop on her current tour. Nokia is the infrastructure vendor
in this market, although we also encountered Corning mmWave small cells within State
Farm stadium.
Our study included walk tests and stationary tests both inside the stadium and around
the entertainment district. Both venues had a wealth of 5G mmWave radios, plus there
was some Band n77 and Band n2 coverage. Since we could identify the locations of many
sites, we were able to do sensitivity studies that looked at the impact of smartphone
orientation, body blockage, and distance on uplink [and some downlink] performance.
For most of the tests we used a newly purchased Galaxy S23 smartphone. However, for
some comparative tests we used legacy smartphones, including the Galaxy S22 Ultra,
the Galaxy S21 FE, and the Galaxy S20 Plus. A key focus of the comparative tests was
the impact of uplink data transfers on battery life (current drain). We also looked at
the impact of 5G mmWave data transfers on battery temperature, including our patent
pending test methodology dubbed the “ice cube” effect.
We can best describe 4CC uplink performance as being “temperamental,” but we attri-
bute this viewpoint, at least in part, to an initial network configuration that didn’t seem
optimized for 4CC uplink. Following some network configuration changes, 4CC uplink
was more prevalent in our tests. However, the impact of smartphone orientation and
body blockage was much greater than we are accustomed to seeing.
Consistent with earlier studies involving current efficiency, or the transferred data speed
relative to the required battery current drain, we found that while higher uplink data
speeds increased the current drain, the higher data speeds more than offset the increase
in battery current. The Galaxy S23 smartphone had substantially better current efficiency
(Mbps/mA) than some of the legacy smartphones we tested. We surmise the improve-
ment in current efficiency could be due, in part, to improvements in the underlying 5G
baseband modems used in these phones.
Lastly, we found that while downlink/uplink data transfers using 5G mmWave had an
impact on battery temperature, the biggest culprit, by far, was the glare from the sun.
This conclusion doesn’t make things any better for 5G smartphone users who reside in
locations where there is sunshine, not to mention relatively warm temperatures.
A special thanks to Accuver Americas (XCAL-Solo, XCAL5, and XCAP) and Spirent
Communications (Umetrix Data) for the use of their respective test equipment and plat-
forms. Both companies have been valued partners for more than 15 years.
3 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

When Verizon first launched 5G mmWave in April 2019, the network was limited to four 100 MHz
component carriers (4CC) in the downlink direction while all the uplink data traffic went over LTE.
The latter attribute didn’t stop “industry pundits” from posting Speedtest results and claiming both
the downlink and uplink speeds were due to 5G mmWave. In June 2020, we published results from
a benchmark study that showed uplink data transfers actually using 5G mmWave (SA 06/11/20,
“What Goes Down, Can Finally Go Up”). The peak speed at the physical layer (PUSCH) was
limited to 98 Mbps with an average speed of 31.5 Mbps. The network at the time was limited to a
single 100 MHz uplink channel.
Fast forward to 2021, we visited Tokyo, Japan to test 5G mmWave (28 GHz) and there we got
the chance to kick the tires of 2CC uplink over 5G mmWave. We documented peak uplink speeds
at or slightly above 300 Mbps with average speeds just under 100 Mbps. Further, we witnessed
wide availability of uplink 2CC, with distances extending to approximately 200 meters from the
serving cell site. In essence, if the smartphone had 1CC 5G mmWave uplink then it most likely
had 2CC uplink.
For the last few months 4CC 5G mmWave uplink has existed in some operator networks but we
just had the opportunity to evaluate its performance. We reached out to AT&T in March, and they
steered us to Glendale, AZ to do the study. AT&T provided logistical support, but the operator and
no other company had any financial investment in this endeavor. As a courtesy, we provided AT&T
with a pre-brief just prior to publishing this report.
Taking a step back, we revisited some of the results from our FR1 device + chipset benchmark The results for the Motorola
study that we published in December (SA 12/07/22, “Cage Match: (FR1 in the Wild!”). We now edge (2022) smartphone,
believe the results for the Motorola edge (2022) smartphone with the MediaTek Dimensity 1050 which we published in an
earlier Signals Ahead report,
chipset do not reflect the smartphone’s actual performance capabilities. For unknown reasons, we
do not reflect its actual
suspect the concurrent logging of the chipset data had a substantial impact on its performance performance capabilities.
and that the published results understate the smartphone’s capabilities. We haven’t experienced this
issue with other entry-level smartphones, including Motorola smartphones, or with the MediaTek
chipset, so the issue could be specific to this model or to the phone we used for the study. Since we
don’t have valid test results for this smartphone, we currently have no viewpoint on its performance,
other than a strong belief that its true performance can be much better than shown in that report.
We kindly remind our subscribers that Signals Ahead is a subscription-based service. This bench- We kindly remind our readers
mark study, like all our studies published in Signals Ahead, is entirely supported by these subscrip- that Signals Ahead is a
tions. We expect our subscribers to respect the amount of effort and expense associated with doing subscription-based service
and that you must abide by
these studies. Although companies with a global license may share these reports internally within
the terms of the subscription.
their organization, we do not allow any external sharing of this report, either in whole or in part.
Companies with a group subscription cannot share the report, either in whole or in part, beyond the
individuals signed up to receive these reports. If you feel compelled to share these reports, then you
should also feel compelled to upgrade your company’s subscription. If you are not sure what type
of subscription your company has then please reach out and ask us. We also note that we provide a
report preview of each Signals Ahead report. You can share the report preview, which is available for
download on our website or by contacting us.
When we first started testing on Day 1, we had a lot of difficulty with our smartphone using
4CC uplink, even though we were pushing full buffer data transfers and seeing reasonably good
data speeds in the downlink direction. This testing included the inside of State Farm stadium. The
next morning, we observed 4CC uplink, but its availability wasn’t what we had expected so we
reached out to AT&T. The operator made a few changes to its network, and we repeated our walk
test through a meaningful cluster of 5G mmWave sites that surrounded the entertainment district.
In the main body of this report, we discuss these changes and show their impact on uplink perfor-
mance. We also discuss why some of the mmWave radios within the stadium were limited to 400
MHz in the downlink even though the operator had 600 MHz of spectrum.

Suffice it to say that with the changes we saw a marked improvement in the availability of 4CC, The 5G mmWave uplink
not to mention overall uplink data speeds. However, we would still describe the 5G mmWave uplink performance was a bit
performance as being a bit temperamental with phone orientation and body blockage having a much temperamental with
phone orientation and
bigger impact on performance than we are accustomed to seeing. Uplink performance with ideal
body blockage having a
conditions and while stationary also delivered speeds that were lower than anticipated. However, much bigger impact on
we partially attribute this outcome to the downlink/uplink ratio used in the network with more performance than we are
bandwidth (time) allocated to the downlink direction than used in other networks. With TDD, the accustomed to seeing.
channel is used for both downlink and uplink data transfers so an increase in the time allocated to
the downlink direction impacts the time available for uplink data transfers, and vice versa. This is a
critical point that readers need to consider when comparing results across networks, especially since
social media and press releases either don’t include this level of detail (social media) or they gloss over
it (press releases). Given that most data traffic is in the downlink direction, the network in Glendale
“only” had 600 MHz of spectrum, and there was 4CC in the uplink direction, it makes sense for a
more downlink centric configuration. We discuss the conditions under which AT&T might change
the downlink/uplink ratio in the main body of this report.
Our battery temperature and battery current tests delivered the expected results. We successfully
transferred far more data over 5G mmWave in a downlink test than most people consume in an
entire month without the smartphone reaching its thermal limit. We achieved the same outcome
in our uplink test with 4CC. Both tests used the Galaxy S23 smartphone. The critical caveat is that
we did these tests when the sun was setting. With sunlight and a reasonably warm day (temps in the
low 80°s F), a phone can reach its 5G thermal limit before downloading/uploading a token amount
of data.
The battery current test results were stellar for uplink 4CC and without any caveats. As we have The battery current test
observed in the past, when a smartphone receives/sends data at a faster rate it generates more current. results were stellar for uplink
However, we also observed in this recent study that the increase in battery current drain was more 4CC and without any caveats.
than offset by the increase in data speeds. What matters is the current efficiency, which we define
as the amount of current required to achieve a given data rate (Mbps/mA). Our results also take
into consideration and remove the effect of the display backlight. This action is critical because
phones come in different shapes and sizes, plus it is difficult to exactly match the display brightness
on each phone. We also point out that since the backlight display is frequently illuminated when
downloading/uploading data, a slower data speed subsequently increases the amount of time the
phone’s display is turned on. This situation further reduces the effective life of the battery – a key
consideration that we intentionally excluded from our analysis since it is difficult to objectively
quantify.
Our detailed analysis of 5G mmWave 4CC uplink performance includes the following:
➤ Comparative results and detailed analysis for most pertinent KPIs between two network
configurations
➤ KPIs analyzed include:
♦ Distribution of uplink throughput by component carrier
♦ Uplink throughput (PUSCH), MCS, RB, PUSCH transmit power versus RSRP
♦ Uplink throughput versus PUSCH transmit power
♦ PUSCH transmit power versus RSRP
♦ Impact of smartphone orientation and body blockage on RSRP and throughput
➤ Mapping out the 39GHz performance in State Farm stadium
➤ Comparative results on the Verizon 28 GHz network

➤ The impact of downlink/uplink data transfers on battery temperature
➤ The impact of 4CC uplink data transfers on battery life, including comparative results for legacy
Samsung flagship smartphones going back to the Galaxy S20 Plus
Chapter 2 discusses the key observations from this study. Chapter 3 provides detailed results
and analysis. Chapter 4 provides our test methodology and Chapter 5 concludes with some final
thoughts.

Table of Contents
1.0 Executive Summary………………………………………………………………………………………………………………………………………… 3
2.0 Key Observations…………………………………………………………………………………………………………………………………………… 11
3.0 Detailed Results and Analysis…………………………………………………………………………………………………………………………… 15
3.1 Before and After……………………………………………………………………………………………………………………………………… 15
3.2 Galaxy S23 and Galaxy S22 Ultra Comparative Results…………………………………………………………………………… 27
3.3 Verizon 5G mmWave Uplink Performance – Sample Results…………………………………………………………………… 30
3.4 4CC Uplink and the Impact on Current Drain……………………………………………………………………………………………… 33
3.5 5G mmWave Data Transfers and the Impact on Battery Temperature………………………………………………………… 41
3.6 5G mmWave Downlink Performance……………………………………………………………………………………………………… 44
3.7 State Farm Stadium Performance Results……………………………………………………………………………………………… 48
4.0 Test Methodology…………………………………………………………………………………………………………………………………………… 52
5.0 Final Thoughts…………………………………………………………………………………………………………………………………………………54
Index of Figures & Tables

Figure 1. 5G mmWave and Band n77 Coverage - Before………………………………………………………………………………………………… 15
Figure 2. 5G mmWave and Band n77 Coverage - After………………………………………………………………………………………………… 16
Figure 3. Average Uplink Throughput – Total and by Component Carrier (Before)…………………………………………………………… 16
Figure 4. Distribution of Uplink Throughput – Total and by Component Carrier (Before)………………………………………………… 17
Figure 5. Average Uplink Throughput – Total and by Component Carrier (After)…………………………………………………………… 17
Figure 6. Distribution of Uplink Throughput – Total and by Component Carrier (After)………………………………………………… 17
Figure 7. Average and Distribution of RSRP –by Component Carrier (Before)………………………………………………………………… 18
Figure 8. Average and Distribution of RSRP – by Component Carrier (Before)………………………………………………………………… 18
Figure 9. Geo Plot of 5G P Cell RSRP (Before)……………………………………………………………………………………………………………… 19
Figure 10. Geo Plot of 5G P Cell RSRP (After)……………………………………………………………………………………………………………… 19
Figure 11. Uplink Throughput Versus RSRP –by component carrier (Before)………………………………………………………………… 20
Figure 12. Uplink Throughput Versus RSRP –by component carrier (After)…………………………………………………………………… 20
Figure 13. Average and Distribution of Uplink PUSCH Transmit Power – by component carrier (Before)…………………………… 21
Figure 14. Average and Distribution of Uplink PUSCH Transmit Power – by component carrier (After)……………………………… 21
Figure 15. PUSCH Transmit Power Versus RSRP –by component carrier (Before)…………………………………………………………… 22
Figure 16. PUSCH Transmit Power Versus RSRP –by component carrier (After)…………………………………………………………… 22
Figure 17. Uplink Throughput Versus PUSCH Transmit Power – by component carrier (Before)……………………………………… 23
Figure 18. Uplink Throughput Versus PUSCH Transmit Power – by component carrier (After)……………………………………… 23
Figure 19. Average and Distribution of Uplink Resource Block Allocations Versus RSRP – by component carrier (Before)… 24
Figure 20. Average and Distribution of Uplink Resource Block Allocations Versus RSRP – by component carrier (After)… 24
Figure 21. Average and Distribution of Uplink MCS Versus RSRP – by component carrier (Before)………………………………… 25

Figure 22. Average and Distribution of Uplink MCS Versus RSRP – by component carrier (After)………………………………… 25
Figure 23. Distribution of 5G and Component Carrier Usage……………………………………………………………………………………… 25
Figure 24. Transmission Type Versus RSRP (After)……………………………………………………………………………………………………… 26
Figure 25. 5G mmWave Usage – 1CC and 4CC (S23 versus S22 Ultra)…………………………………………………………………………… 27
Figure 26. P Cell PUSCH Throughput Versus RSRP – Galaxy S23 versus Galaxy S22 Ultra……………………………………………… 27
Figure 27. LTE and 5G PUSCH Throughput Versus RSRP Time Series Plot……………………………………………………………………… 28
Figure 28. P Cell RSRP and the Impact of Body Blockage and Phone Orientation………………………………………………………… 29
Figure 29. PUSCH Throughput – Total and by Component Carrier (Verizon)………………………………………………………………… 30
Figure 30. RSRP Time Series - by Component Carrier (Sensitivity Study)………………………………………………………………………… 31
Figure 31. Average RSRP - by Component Carrier (Sensitivity Study)……………………………………………………………………………… 31
Figure 32. PUSCH Throughput Time Series - by Component Carrier (Sensitivity Study)……………………………………………………32
Figure 33. Average PUSCH Throughput - by Component Carrier (Sensitivity Study)…………………………………………………………32
Figure 34. Battery Current Drain in Airplane Mode……………………………………………………………………………………………………… 33
Figure 35. Uplink Application Layer Throughput after Dark – by smartphone……………………………………………………………… 34
Figure 36. Battery Current Drain During Uplink Data Transfer after Dark Time Series – by smartphone………………………… 35
Figure 37. Average Adjusted Battery Current Drain During Uplink Data Transfer after Dark – by smartphone………………… 35
Figure 38. 5G mmWave Uplink Data Transfer Current Efficiency after Dark – by smartphone……………………………………… 36
Figure 39. View of the 5G mmWave Radio at Lunch…………………………………………………………………………………………………… 36
Figure 40. Uplink Application Layer Throughput after Lunch Time Series – by smartphone……………………………………………37
Figure 41. Average Uplink Application Layer Throughput after Lunch – by smartphone……………………………………………………37
Figure 42. Average Adjusted Battery Current Drain During Uplink Data Transfer after Lunch – by smartphone……………… 38
Figure 43. 5G mmWave Uplink Data Transfer Current Efficiency after Lunch – by smartphone……………………………………… 38
Figure 44. Uplink Application Layer Throughput on Bench – by smartphone……………………………………………………………… 39
Figure 45. Average Adjusted Battery Current Drain During Uplink Data Transfer on Bench – by smartphone………………… 39
Figure 46. 5G mmWave Uplink Data Transfer Current Efficiency on Bench – by smartphone………………………………………… 40
Figure 47. Taking a Refreshing Ice Bath ……………………………………………………………………………………………………………………… 41
Figure 48. 5G mmWave Uplink Data Transfers and the “Ice Cube” Effect…………………………………………………………………… 42
Figure 49. 5G mmWave Downlink Data Transfers……………………………………………………………………………………………………… 43
Figure 50. 5G mmWave Uplink Data Transfers…………………………………………………………………………………………………………… 43
Figure 51. 5G mmWave Downlink Walk Test……………………………………………………………………………………………………………… 44
Figure 52. P Cell 5G mmWave PDSCH Throughput Versus P Cell SINR………………………………………………………………………… 45
Figure 53. Total 5G mmWave PDSCH Throughput Versus P Cell SINR………………………………………………………………………… 45
Figure 54. P Cell RSRP Versus Cell Site Distance – PCI 65…………………………………………………………………………………………… 46
Figure 55. Total 5G mmWave PDSCH Throughput Versus Cell Site Distance – PCI 65…………………………………………………… 46
Figure 56. Total 5G mmWave PDSCH Throughput Geo Plot……………………………………………………………………………………… 47
Figure 57. Stadium Map PCI Values…………………………………………………………………………………………………………………………… 48

Figure 58. Stadium Map 5G Downlink Throughput…………………………………………………………………………………………………… 49
Figure 59. 5G P Cell SINR by PCI……………………………………………………………………………………………………………………………… 50
Figure 60. 5G Total Throughput by PCI…………………………………………………………………………………………………………………… 50
Figure 61. 5G P Cell RSRP by PCI………………………………………………………………………………………………………………………………… 51
Figure 62. Umetrix Data Platform………………………………………………………………………………………………………………………………52
Figure 63. XCAL-Solo…………………………………………………………………………………………………………………………………………………53

INCASE
IN CASEYOU
YOU MISSED
MISSED IT:
IT:
SIGNALS
SIGNALS AHEAD
AHEAD BACK
BACK ISSUES
ISSUES
➤ 03/23/23 "5G: The Greatest Show on Earth! Vol 31: ➤ 1/10/23 “5G: The Greatest Show on Earth! Vol 30:
This is a Job for HPUE!” "SRG just completed its 31st 5G MU-MIMO and the Tower of Power” SRG just completed
benchmark study. For this endeavor we collaborated with Accuver its 30th 5G benchmark study. For this endeavor we collaborated
Americas and Spirent Communications to conduct an indepen- with Accuver Americas and Spirent Communications to conduct
dent benchmark study of a PC 1.5 (power class 29 dBm) versus a an independent benchmark study of 5G 8-layer MU-MIMO,
PC 2 (26 dBm) capable smartphone. using the SRS-based implementation.
Highlights of the Report include the following: Highlights of the Report include the following:
Our Thanks. We did this study in collaboration with Accuver Our Thanks. We did this study in collaboration with Accuver
Americas (XCAL5 and XCAP) and Spirent Communications Americas (XCAL5 and XCAP) and Spirent Communications
(Umetrix Data). SRG is responsible for the data collection and all (Umetrix Data). SRG is responsible for the data collection and all
analysis and commentary provided in this report. analysis and commentary provided in this report.
Background. At 3GPP RAN#78 back in December 2017, a few Our Methodology. Testing took place on the T-Mobile network
companies proposed a higher class power amplifier (29 dBm) for (Band n41) in southern California at commercial cell sites. We
certain LTE and 5G mid-band frequencies to offset the coverage used 4 smartphones or 4 FWA CPEs to load the network with
challenges associated with the higher frequency TDD spectrum. full buffer data transfers. We looked at the impact of UE place-
Five years later, this proposal is finally becoming a commercial ment within the cell as well as mobility. We analyzed all the
reality. We've already tested PC 2 versus PC 3 in an earlier Signals typical KPIs, including RB usage, MIMO layers, MCS, and, of
Ahead report so it was only natural for us to take on this topic. course, throughput, while also including vehicular speed and geo
Our Methodology. Testing took place on the T-Mobile network coordinates.
(Band n41) in and around Laguna Beach, CA. We used 2 largely The Results. We observed significant double-digit throughput
identical Motorola edge (2022) smartphones with the only differ- gains due to MU-MIMO pairing relative to SU-MIMO (we
ence being one phone supported PC 1.5 and one phone supported disabled SRS / MU-MIMO in the network). Close placement of
PC 2. We did downlink and uplink tests, stationary and mobile, UEs had little, if any, impact on the efficiency of MU-MIMO with
as well as individual versus parallel tests. T-Mobile only provided excellent pairing maintained.
logistical support, including the provisioned phones. The FWA Implications. T-Mobile has already deployed the func-
The Results. As expected, the PC 1.5-enabled smartphone tionality at all Ericsson Band n41 cell sites on a nationwide basis.
had higher downlink throughput in RF challenged areas, and it For reasons discussed in the report, MU-MIMO functionality can
provided better coverage, higher uplink MCS values, used more have a significant positive influence on the FWA business case,
PUSCH resource blocks, and made better use of UL-MIMO and even though some limitations to MU-MIMO exist.
UL-256QAM than the PC 2-enabled smartphone. We quantify More in Store. This MU-MIMO report marks what we anticipate
the differences in the report. will be at least a few more MU-MIMO studies in the coming year.
UL-256QAM and UL-MIMO Revisited. We remind readers how We anticipate looking at 16-layer MU-MIMO, more device place-
these two critical features can provide much higher throughput / ment scenarios, different geographies (rural), and traffic profiles.
spectral efficiency than smartphones with only UL-64QAM and All these reports will be available through our Signals Ahead
a single uplink layer. PC 1.5 enhances how frequently these two publication.
features are used. We also compare the Motorola edge (2022)
smartphone's uplink performance with a best-in-class smartphone
that only supported UL-256QAM.
UL-MU-MIMO. Not to confuse readers, but when testing single user
UL-MIMO we discovered uplink Mulit-user MIMO in which
the network supported up to four smartphones, each capable of
concurrently using all possible uplink RBs with certain caveats. We
didn't quantify its benefits, but we plan to do so in an upcoming
report.

2.0 Key Observations
We offer the following key observations based on our analysis of the data.
5G mmWave 4CC uplink can deliver meaningful capacity and user data speeds,
but its performance and availability could be more consistent. In our initial tests,
we had a lot of difficulty getting our Galaxy S23 smartphone to use 5G mmWave 4CC uplink. We
reached out to AT&T, and they made two important changes to their network. First, the operator
removed the data buffer requirement which was used to trigger the use of 5G mmWave. Initially,
the triggering of 5G mmWave was based on the amount of buffered data in the scheduler, but since
our tests were largely uplink centric, our data transfers didn’t trigger the use of mmWave radio
assets. We’ve experienced this same issue in the past on the Verizon network. The second change
adjusted the PUSCH transmit power more evenly across the four component carriers. This change
had a profound impact on the contributions from the three secondary carriers and it meaningfully
increased overall uplink data speeds.
Focusing just on the results after the network changes – referred to as the “After” scenario in this
report, we were still disappointed with the results, especially the consistency and reliability of the
uplink performance. On a consistent basis, we found that phone orientation and body blockage had
a profound impact on uplink performance. The uplink throughput while walking by a 5G mmWave
radio that was a stone’s throw away was extremely low or even nonexistent. When we stopped and
faced the mmWave radio the speeds jumped immediately to speeds that were closer to expectation.
This outcome occurred repeatedly throughout our stay and with all the sites we encountered.
Facing away from the radio and walking away from the serving cell site resulted in a similar
outcome. If we then stopped and faced the radio, the uplink data speeds returned. It was like clock-
work. Looking at some of the underlying data, there could easily be a 15-20 dB difference in the
signal strength (RSRP) from the same location, based on the phone orientation and body blockage.
We held the phone in portrait mode, much like someone would hold the phone while live streaming.
Perhaps holding the phone slightly differently would have produced better results, but it wasn’t as if
we didn’t get good results on occasion. We also didn’t change the position of our hand when we faced
the opposite direction from the cell site when the signal strength dropped.
We don’t know if our experience and the results we obtained are highly influenced by 39 GHz –
most of our earlier testing was using 28 GHz or 26 GHz – or if other factors are at play. If there was
a frequency component at play, then it does call into question the use of higher frequency bands for
mobile data usage. Nonetheless, if a standalone mmWave network is going to provide a compelling
and consistent user experience for both downlink and uplink use cases then there is some work ahead
remaining. 5G mmWave 4CC uplink for fixed wireless use cases could still be compelling since the
CPE is fixed and pointed directly at the serving cell site, but to date we have not had the opportunity
to confirm this hypothesis for ourselves.
NR-DC is [hopefully] the answer. From our testing of 5G mmWave around the world – three
continents and counting – we’ve observed that when a smartphone uses 5G mmWave, it tends to not
receive any data over LTE. This strategy makes sense because operators want to push the data traffic
from their relatively low capacity LTE networks to their high capacity 5G mmWave networks. The
downside is that if all the data traffic is going over 5G mmWave then the reliability and consistency
of the downlink/uplink data transfers hinge on 5G mmWave. Even with the best of circumstances,
5G mmWave can’t deliver the same reliability and consistency that is possible with LTE.
In theory, LTE should pick up at least some of the slack when the 5G mmWave throughput
disappears or reaches unacceptably low levels. From our experiences, including most recently in
Arizona, this transition happens much slower than desired, if at all. In some of our tests from within
the stadium, the 5G mmWave uplink throughput was only in the low tens of Mbps, yet LTE data
traffic was absent. LTE was serving as an anchor cell in the NSA network. And in our outdoor

walk tests, if we lost 5G mmWave uplink data traffic, the transition to Band n77 or LTE only didn’t
occur instantaneously. Tens of Mbps may be considered good over LTE, but it isn’t good over 5G
mmWave and we’re pretty sure the current efficiency with these low data speeds over 5G mmWave
wouldn’t be good.
NR-DC, in which the smartphone is simultaneously receiving or transmitting user plane data NR-DC would deliver a user
over 5G mmWave and a sub 6 GHz channel could be the solution. Today, most mid-band 5G experience that is more
networks still have ample capacity available so leveraging the lower 5G channel for data traffic while commensurate with 5G.
also pushing the preponderance of the data traffic over 5G mmWave wouldn’t penalize mid-band
network performance. The upside benefit is that NR-DC would deliver a user experience that is more
commensurate with 5G. When 5G mmWave underperforms, the mid-band 5G channel would pick
up the slack. The speeds may not equate to what is possible over 5G mmWave, but the throughput
would be much better than what we experience today in the absence of NR-DC. More importantly,
the transition time when the smartphone moved between the FR2 and FR1 networks would be
eliminated since the smartphone would already be using both networks from the start. We look
forward to testing NR-DC later this year.
The higher uplink data speeds can result in a significant increase in battery
current efficiency. We tested four flagship Samsung Galaxy series smartphones, going back
to the Galaxy S20 Plus and up to the most recent Galaxy S23 smartphone. For these tests, we used
a 4-minute or 90-second Umetrix full buffer HTTP uplink data transfer to measure the uplink
throughput of each smartphone. We then looked at application layer throughput and the corre-
sponding current drain of the smartphone to determine each smartphone’s current efficiency, or the
achieved throughput for a given current drain (Mbps/mA). We explain our methodology later in
Section 3.4 and in the Test Methodology chapter.
The Galaxy S23 smartphone had substantially higher current efficiency compared with the other At one test location,
smartphones. For example, at one test location, the Galaxy S23 current efficiency was 8.2x higher the Galaxy S23 current
than the Galaxy S20 Ultra, 5.4x higher than the Galaxy S21 FE and 1.3x higher than the Galaxy efficiency was 8.2x higher
than the S20 Ultra, 5.4x
S22. This strong performance was due to a combination of higher uplink throughput and lower
higher than the S21 FE, and
current drain. We do note, however, that in this test the Galaxy 22, which also supports uplink 1.3x higher than the S22.
4CC, delivered slightly higher uplink throughput but the higher throughput was offset by higher
current drain, compared with the Galaxy S23. Although the results varied across the tests, the
results were directionally similar to the test results we highlight in this chapter. We also note we
adjusted the results to take into consideration differences in the current drain due to the backlight
display to ensure a fair comparison across smartphones.
The results from this study are consistent with what we have witnessed in earlier testing that
we have done. In general terms, when a smartphone receives or transmits data at a faster speed,
the current efficiency increases even though the current drain is higher. In essence, the higher
data current drain is more than offset by the higher data speed. One important caveat is that 5G
mmWave will most likely not have great current efficiency when the data speeds are abnormally low,
for example due to the requirements of the application (video chat) or to poor RF conditions. We
tried to test 5G mmWave uplink with poor conditions for this study, but the legacy smartphones had
difficulty remaining on 5G mmWave, so we abandoned the objective.
Although heavy 5G mmWave usage has an impact on battery temperature, in

our tests the biggest culprit was Mother Nature and the glaring sun over-
head. When we tested outdoors during the mid-day hours (temps in the high 70°s or potentially
low 80°s F), our smartphones got warm/hot relatively quickly. In one test, the Galaxy S21 FE hit the
5G thermal limit (~42° C) about halfway through a 4 minute uplink test, even though we had partial
umbrella coverage, and we hadn’t used the phone earlier in the day. One could easily blame 5G, but

when we reviewed the data log, we discovered the phone’s battery temperature was approaching 40°
C before we even started the test. You can’t blame 5G for that starting temperature.
When we tested in the early evening hours and after the sun had set, it was a whole new ball game
even though the outside temperature had only dropped a degree or two. During an ~18 minute
downlink walk test, we downloaded 170 GB over 5G, largely mmWave but some n77. During this
test the battery temperature increased by 6.7° C to 40.1° C, or just a degree or two below the thermal
threshold. Granted, what saved us and the phone in this test was the starting temperature, but
clearly no one in their right mind tries to download 170 GB in one sitting. We successfully did and
the phone didn’t hit the thermal limit.
In a separate uplink test we transferred 9.4 GB of data over a period of ~18 minutes and the battery
temperature only increased by 5.9° C while remaining at or below 39° C. Had we done this same test
a few hours earlier, we’re almost certain the phone would have hit its thermal limit well before 18
minutes. In fact, we’re pretty certain the phone would have hit its thermal limit if we just held it in
our hand with the screen facing the sun.
We believe operators will continue to adjust the ratio of downlink versus

uplink bandwidth with their mmWave spectrum to meet changing traffic
profiles. The uplink data speeds we observed in Arizona were notably lower than we’ve seen in
other tests on a per 100 MHz carrier basis. However, one important consideration is the ratio of
time slots allocated to downlink versus uplink data traffic. The comparative results for Verizon that
we document in Section 3.3 show much higher uplink throughput for 5G mmWave 4CC, not to
mention more consistency in uplink throughput.
Setting aside the consistency comparison, we can attribute some of the differences in the uplink
throughput between the two networks to differences in the downlink/uplink ratios used by the two
networks. In the case of the Verizon network there were ~23% more uplink resource blocks available
so we’d expect a comparable increase in higher data speeds. Additionally, the allocation of more
uplink resources in the Verizon network impacted the number of downlink resources, which, in
turn, reduced its downlink data speeds over what would be possible with more downlink resources
available. You can’t have your cake and eat it too.
In a social media post that we observed, someone from Verizon was highlighting its 5G mmWave
performance in Kansas City, where the NFL draft was recently held. The screenshot showed an
average uplink data speed of just over 900 Mbps, which is awfully darn good. However, the poster
also noted Verizon had allocated more bandwidth to the uplink direction – presumably, because
attendees at the NFL draft like to stream video from the event.
Since 5G mmWave deployments are generally isolated from each other, it is relatively straight
forward to adjust the downlink/uplink ratios by venue and/or event to accommodate expected traffic
profiles. This action is far more difficult with mid-band spectrum since the change would need to
occur at a market level. For these reasons, we expect operators will begin experimenting with how
they partition their TDD 5G mmWave spectrum.
AT&T indicated it may look at adjusting its downlink/uplink ratio, but only after first making
sure it has optimized its uplink performance. In the case of Arizona, the operator “only” has 600
MHz of spectrum with some sites currently limited to 400 MHz. The 400 MHz limitation is due to
the current software which doesn’t scale to support both 600 MHz of spectrum and a large number
of connected devices. Therefore, mmWave radios covering a large section of the stadium seating were
artificially limited to 400 MHz to ensure they could support the anticipated number of connected
devices. Our seat during the game fell into this category.
Since data traffic is almost always skewed to the downlink direction and since AT&T has launched
4CC for the uplink, there is perhaps a greater need to preserve the higher allocation of downlink
network resources, at least for the time being. As operators allocate more spectrum to the uplink

direction it will be critical that they ensure the network performance meets user expectations and
that the data speeds are commensurate with 5G mmWave.

3.0 Detailed Results and Analysis
In this chapter we provide results and analysis of all our testing in Glendale. The first section provides
comparative results for the uplink performance based on the initial network parameter settings and
the results following two changes in the network configuration. We also include comparative results
for the Galaxy S22 Ultra and Galaxy S23 smartphones (Section 3.2), and Verizon 28 GHz (Section
3.3). Section 3.4 provides results from our testing of uplink performance and its impact on battery
current drain while Section 3.5 focuses on 5G mmWave and its impact on the battery temperature.
In Section 3.6 we look at 39 GHz downlink performance and in the last section we show results
from testing inside State Farm stadium.
3.1 Before and After

After our first day of testing and some initial walk testing on our second day in the desert, AT&T
made two changes to its network configuration. First, it removed the RLC buffer restriction for
how the network triggers 5G mmWave for uplink data transfers. Initially, the downlink buffer was
used to trigger 5G mmWave but since our data transfers were entirely in the uplink direction, 5G
mmWave rarely got triggered – even though we were using full buffer data transfers. Secondly,
AT&T modified the distribution of PUSCH transmit power between the four component carriers
so that it was more evenly distributed. Initially, the transmit power was skewed to the P Cell at the
expense of the three secondary cells. There are pros and cons to these network settings, but from the
perspective of our specific use case (uplink centric), the results were much better with the “After”
scenario. Because we think it is interesting, and because the “Before” results reflect how the network
performed prior to our arrival, we are showing both sets of results in this section. Thankfully, Taylor
Swift doesn’t want anyone live streaming her concerts.
Figure 1 (Before) and Figure 2 (After) show the walk routes we used for these tests as well as
showing where we encountered each 5G network. Although the After image shows the S23 rarely
used n77 when not using mmWave, we do not believe it is a meaningful observation since changes
to mmWave settings would have no bearing on n77 coverage. Further, in other tests the smartphone
used n77 in these areas. However, it is possible the differences in mmWave usage are valid or at least
it is plausible that lower P Cell PUSCH transmit power could influence the smartphone’s use of
mmWave for uplink transmissions. Then again, we note the RSRP between the two walk tests was
very similar, as shown in subsequent figures in this section.
Figure 1. 5G mmWave and Band n77 Coverage - Before

No 5G
Band n77
5G mmWave
Source: Signals Research Group

Figure 2. 5G mmWave and Band n77 Coverage - After
No 5G
Band n77
5G mmWave
Cumulative Probability Distribution (%)

Figure 3 (average values) and
100% Figure 4 (distribution) show the PUSCH throughput for each With the Before scenario,
component carrier as well as the total uplink throughput for the Before scenario. Figure 5 and the smartphone only used
Figure 6 show the same informationCumulative
for the After scenario. Since(%)
the Galaxy S23 smartphone used 4CC for 6% of the time,
80% Probability Distribution
increasing to 61% of the time
4CC to a much greater extent with100% the After scenario, the total PUSCH throughput was nearly
with the After scenario.
twice as high as the Before scenario.
75%
With the Before scenario, the smartphone only used 4CC for
60% P Cell
6% of the time; it used 1CC for 38% of the time. With the After scenario, the Galaxy S23 increased
S1 Cell
50% S2 Cell
its use of 4CC to 61% of the time while it used 1CC for 66% of the time. At the same S3 Celltime, the P
40%
Cell throughput was higher with the25%Before scenario (27 Mbps) than with the After scenario (18.5
Mbps). Since these percentages are 0%
not geo binned and since we didn’t exactly follow the same route
20%
each time, these percentages are not precise
0 for comparison
20 purposes.
40 Nonetheless,
60 80 it is clear
100 4CC
usage was much higher with the After scenario. Mbps
0%
0 50 100 150 200 250 300 350 400
Figure 3. Average Uplink Throughput – TotalMbps
and by Component Carrier (Before)
35.3
27.0
2.8 2.7 2.8
P Cell S1 Cell S2 Cell S3 Cell Total

Average (Mbps)

Figure 4. Distribution of Uplink Throughput – Total and by Component Carrier (Before)

100%
Cumulative
80% Probability Distribution (%) Cumulative Probability Distribution (%)
100% 100%
60% 75%
P Cell
80% S1 Cell
50% S2 Cell
Cumulative Probability Distribution (%) S3 Cell
40% 100%
60% 25%
75% P Cell
0% S1 Cell
20% 50% S2 Cell
0 20 40 60 80 S3 Cell 100
40%
25% Mbps
0%
20% 0%
0 50 100 150 200 250 300 350 400
0 20 40 60 80 100
Mbps Mbps
0%
0 50 100 150 200 250 300 35.3
350 400
Mbps
Figure 5. Average Uplink Throughput – Total and by Component Carrier (After)
27.0
71.1
2.8 2.7 2.8
P Cell
18.5 S1 Cell
18.2 S2 Cell S3 Cell Total
17.3 17.1
Average (Mbps)
P Cell S1 Cell S2 Cell S3 Cell Total

Average (Mbps) Source: Signals Research Group
Figure 6. Distribution of Uplink Throughput – Total and by Component Carrier (After)

100%
80%
100%
60%
75% P Cell
S1 Cell
50% S2 Cell
S3 Cell
40%
25%
20% 0%
0 20 40 60 80 100
Mbps
0%
0 50 100 150 200 250 300 350 400
Mbps
17 71.1 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

As just indicated, despite the big differences in uplink performance, the mmWave signal strength
(RSRP) was comparable in the two tests. This information is reflected in Figure 7 and Figure 8.
Figure 7. Average and Distribution of RSRP –by Component Carrier (Before)

100% P Cell S1 Cell S2 Cell S3 Cell
S1 Cell
80%
S3 Cell P Cell
60%
S2 Cell
40%
20%
−94.8
0% −103.0 −102.2 −102.2
−120 −110 −100 −90 −80 −70
RSRP (dBm) Average RSRP (dBm)
Figure 8. Average and Distribution of RSRP – by Component Carrier (Before)

P Cell S1 Cell S2 Cell S3 Cell
100%
S1 Cell
80%
P Cell
60%
S3 Cell
40%
20%
S2 Cell
−94.5
0%
−102.7 −102.1 −102.0
−120 −110 −100 −90 −80 −70
RSRP (dBm) Average RSRP (dBm)

While walking by an adjacent
Figure 9 (Before) and Figure 10 (After) provide geo plots of the same information, albeit limited mmWave radio, we turned
to just the P cell. In Figure 9, the inset and the two red rectangles in the main figure highlight to face the site at which
point there was a dramatic
areas where there was a significant increase in the RSRP. Although these two places were located
increase in RSRP as well
immediately in front of the serving cell site, we attribute the sharp spike in signal strength to the as uplink throughput.
orientation of the phone. Specifically, when walking by the cell site we turned to face the adjacent
mmWave radio at which point there was a dramatic increase in RSRP and a corresponding increase
in the uplink throughput. We revisit this topic throughout this chapter.
Figure 9. Geo Plot of 5G P Cell RSRP (Before)
RSRP (dBm)
= Band n77
= > -75
= > -80
= > -85
= > -90
= > -95
= > -100
= < -100
Figure 10. Geo Plot of 5G P Cell RSRP (After)

RSRP (dBm)
= Band n77
= > -75
= > -80
= > -85
= > -90
= > -95
= > -100
= < -100

Figure 11 (Before) and Figure 12 (After) show the relationship between the uplink throughput for
each component carrier relative to that carrier’s signals strength. In the Before scenario, there was
only uplink throughput on the secondary carriers when the RSRP was higher than -95 dBm. With
the After scenario, the range extended down to -110 dBm. With the more unbalanced P cell power
level applied in the Before scenario, the P Cell uplink throughput was higher than the secondary
cells. This outcome was reversed with the After scenario.
Figure 11. Uplink Throughput Versus RSRP –by component carrier (Before)
PUSCH Uplink Throughput (Mbps)
100
S1 Cell
S3 Cell
80 P Cell
60
40
20
S2 Cell
0
−110 −105 −100 −95 −90 −85 −80 −75 −70
RSRP (dBm)
Figure 12. Uplink Throughput Versus RSRP –by component carrier (After)
PUSCH Uplink Throughput (Mbps)
100
80
60
S3 Cell S2 Cell
S1 Cell
40
P Cell
20
0
−110 −105 −100 −95 −90 −85 −80 −75 −70
RSRP (dBm)

In Figure 13 and Figure 14 we illustrate the average PUSCH transmit power for each component
carrier as well as each carrier’s distribution of power transmit levels. It is evident that with the
Before scenario the P cell transmit power level was higher than it was with the secondary cells
while with the Average scenario the transmit power levels were more balanced across the P cell and
secondary carriers. It isn’t clear, however, why the average power levels dropped so substantially
with the After scenario.
Figure 13. Average and Distribution of Uplink PUSCH Transmit Power – by component carrier (Before)
100% 9.4
S2 Cell
80%
S1 Cell 7.0 7.0 7.1
60%
40%
S3 Cell
20%
P Cell
0%
−5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 P Cell S1 Cell S2 Cell S3 Cell
PUSCH Transmit Power (dBm) Average (dBm)
Figure 14. Average and Distribution of Uplink PUSCH Transmit Power – by component carrier (After)
100%
S2 Cell
P Cell
80%
S3 Cell
60%
0.1
−0.2 −0.1 −0.2
40% S1 Cell
20%
0%
P Cell S1 Cell S2 Cell S3 Cell
−5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
PUSCH Transmit Power (dBm) Average (dBm)

Figure 15 and Figure 16 provide the relationship between the transmit power level and the
corresponding RSRP for each component carrier. Both figures show the expected drop in transmit
power levels with increasing RSRP. In absolute terms, the transmit power levels were much higher
for all component carriers with the Before scenario. Additionally, with more favorable RSRP there
was a sharper drop in the transmit power levels for the secondary carriers than there was with the
After scenario.
Figure 15. PUSCH Transmit Power Versus RSRP –by component carrier (Before)
PUSCH Tx Power (dBm)
12
10
4 S2 Cell
0
S3 Cell
P Cell
−2
−4 S1 Cell
−110 −105 −100 −95 −90 −85 −80 −75 −70

RSRP (dBm)
Figure 16. PUSCH Transmit Power Versus RSRP –by component carrier (After)
PUSCH Tx Power (dBm)
12
10
−2 S2 Cell
S3 Cell P Cell
−4 S1 Cell
−110 −105 −100 −95 −90 −85 −80 −75 −70

RSRP (dBm)

Next, we show the uplink throughput as a function of the transmit power. Figure 17 provides this
information for the Before scenario and Figure 18 provides this information for the After scenario.
The results in the two figures may be a bit confusing at first since the throughput in the Before
scenario seems to be higher than it was in the After scenario, or the inverse of what we showed in the
earlier figures. However, in the Before scenario the average PUSCH transmit power for each carrier
was approximately 7-9 dB, or the region in Figure 17 where the Before PUSCH throughput was
nonexistent for the secondary carriers. In other words, although the PUSCH throughput for each
component carrier was quite high with lower transmit power and the Before scenario, this situation
rarely occurred in the tests.
Figure 17. Uplink Throughput Versus PUSCH Transmit Power – by component carrier (Before)
PUSCH Throughput (Mbps)
100
P Cell
80
S1 Cell S2 Cell
60
40
20 S3 Cell
0
−5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
PUSCH Transmit Power (dBm)
Figure 18. Uplink Throughput Versus PUSCH Transmit Power – by component carrier (After)
100
80
60
40
S1 Cell
S2 Cell
20
P Cell S3 Cell
0
−5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
PUSCH Transmit Power (dBm)

The next four figures look at uplink performance and potential coverage constraints from a different
perspective. Figure 19 (Before) and Figure 20 (After) provide the average and distribution for the
uplink resource blocks. Figure 21 (Before) and Figure 22 (After) show the average and distribution
for the uplink MCS values.
Figure 19. Average and Distribution of Uplink Resource Block Allocations Versus RSRP – by component carrier (Before)
Uplink PRBs
12
6.5
10
P Cell
6
S1 Cell
4
S2 Cell
2 1.0 1.0 1.0
0 S3 Cell
−110 −105 −100 −95 −90 −85 −80 −75 −70 P Cell S1 Cell S2 Cell S3 Cell
RSRP (dBm) PUSCH RBs (Average)
Figure 20. Average and Distribution of Uplink Resource Block Allocations Versus RSRP – by component carrier (After)
Uplink PRBs
12
7.0 7.1 7.1
10 6.3
S1 Cell
8
S2 Cell S3 Cell
4
P Cell
0
RSRP (dBm) PUSCH RBs (Average)
Both sets of figures show results that are consistent with the PUSCH throughput. With the
balanced power levels, the uplink RB usage was relatively consistent across the four carriers while
with the Before scenario, the uplink RB usage was skewed to the P Cell. The MCS results tell a
similar story. However, it is worth pointing out the MCS value for the P Cell was a bit higher with
the Before scenario than with the After scenario. Additionally, with the After scenario, the MCS
values for the secondary cells was slightly lower than it was for the P Cell.

Figure 21. Average and Distribution of Uplink MCS Versus RSRP – by component carrier (Before)
Uplink MCS
25
12.2
20
P Cell
15
S1 Cell
10
S2 Cell
5
2.1 2.2 2.2
S3 Cell
0
RSRP (dBm) Uplink MCS (Average)
Figure 22. Average and Distribution of Uplink MCS Versus RSRP – by component carrier (After)
Uplink MCS
25
P Cell 11.4
20
9.0 8.7
S1 Cell
15 7.4
S2 Cell
10
5
S3 Cell
0
RSRP (dBm) Uplink MCS (Average)
Figure 23. Distribution of 5G and Component Carrier Usage

LTE Only 0%
LTE Only No Meaningful P Cell
Band n77 Usage Throughput (< 1 Mbps)
21% 17% No Meaningful P Cell
20%
Band n77 Usage 1% Throughput (< 1 Mbps)
35%
1CC (>1 Mbps) 6%
1CC (>1 Mbps)

5G mmWave Usage 5G mmWave Usage 76%
80% 79%
4CC (>1 Mbps per CC)
59%
4CC (>1 Mbps per CC) 7%

Before After Before After
Usage (%) Usage (%)

To conclude this section, Figure 24 shows the distribution of the uplink transmission type as a
function of the RSRP for the After scenario. The figure only shows results for the P Cell and S1 Cell.
We are including this figure because we know someone will ask if we didn’t include it. The three
transmission types shown in the figure are SISO (Single Input, Single Output), SMDL (Spatial
Multiplexing, Dual Layer), and SMSL (Spatial Multiplexing, Single Layer). The results show the
SMDL distributions were more skewed toward the higher RSRP values while SISO/SMSL were
more prevalent with lower RSRP. Both outcomes match what we would expect.
Figure 24. Transmission Type Versus RSRP (After)

100%
S1 Cell SMSL
P Cell SISO
S1 Cell SMDL
80%
60% P Cell SMDL
40%
20%
0%
−110 −105 −100 −95 −90 −85 −80 −75 −70
RSRP (dBm)

3.2 Galaxy S23 and Galaxy S22 Ultra Comparative Results
We did a comparative uplink walk test to compare uplink performance between the Galaxy S23 and
Galaxy S22 Ultra smartphones. These tests used the “After” network configuration. We repeated the
test for the Galaxy S22 Ultra but in both tests the smartphone reverted to 1CC throughout much
of the walk. Thermal issues were not the problem, based on the battery temperature reported by the
smartphone. As shown in a later section, the Galaxy S22 Ultra definitely supported 4CC uplink,
but its use was infrequent in these two walks. Figure 25 shows the percentage of time the two
smartphones used 4CC versus 1CC relative to the total 5G mmWave connection time. Additionally,
the figure shows how frequently each smartphone used 5G mmWave during the test. Although the
S22 Ultra used 1CC more often than the Galaxy S23 (58% versus 51%), the Galaxy S23 used 4CC
far more often when using 5G mmWave than the Galaxy S22 Ultra (97% versus 19%). When doing
this analysis, we required the uplink throughput to be at least 1 Mbps – a pretty low threshold.
Figure 25. 5G mmWave Usage – 1CC and 4CC (S23 versus S22 Ultra)
97%
81%
58%
51% 49%
42%
19%
3%
4CC Uplink Usage 1CC Uplink Usage 4CC Uplink Usage 1CC Uplink Usage 5G mmWave Usage No mmWave 5G mmWave Usage No mmWave
Galaxy S22 Galaxy S23 Galaxy S22 Galaxy S23

As reflected in Figure 26, the S22 Ultra had higher uplink throughput on the 5G mmWave P cell
than the Galaxy S22 smartphone. However, it could very well be the case that the S23 smartphone’s
use of 4CC impacted the throughput it achieved on the P cell so while the information in the figure
is correct, it could be misleading.
Figure 26. P Cell PUSCH Throughput Versus RSRP – Galaxy S23 versus Galaxy S22 Ultra
P Cell PUSCH Uplink Throughput (Mbps)
100
Galaxy S22
80
60
40
Galaxy S23
20
0
−110 −105 −100 −95 −90 −85 −80 −75 −70
RSRP (dBm)

Although this comparative study turned out to not be as useful as we were hoping, the information
in Figure 27 is insightful since it shows how 5G mmWave uplink throughput was impacted during
the walk test. At roughly 20 seconds into the portion of the test shown in the figure, the Galaxy 23
dropped the Band n77 connection and moved to 5G mmWave. However, after establishing a connec-
tion it took another 12 seconds for the uplink data transfer to resume on 5G mmWave. At roughly
60 seconds, the 5G mmWave signal deteriorated and the uplink data transfer over mmWave stopped
with the uplink throughput continuing on LTE. Shortly thereafter, the 5G mmWave RF conditions
improved but the data transfer remained only on LTE – no 5G mmWave. At the 160 time stamp
we put the phone into and out of airplane mode. Upon doing so, the uplink data transfer resumed
on 5G mmWave with no contribution from LTE. The figure also shows that LTE throughput only
occurred when the smartphone was using Band n77 or no 5G whatsoever.
Figure 27. LTE and 5G PUSCH Throughput Versus RSRP Time Series Plot
Mbps 5G P Cell RSRP (dBm)
300 −60
12 second transition from FR1 to FR2 Airplaned Phone
n77 (5G throughput resumed)
with no LTE or 5G throughput
250
−70
Total 5G P Cell RSRP
200
−80
150
5G mmWave RSRP improves but
5G mmWave throughput does not return −90
100
−100
50
LTE PUSCH
0 −110
0 20 40 60 80 100 120 140 160
Time (sec)

Figure 28 shows how body blockage and the orientation of the smartphone relative to the serving
cell site impacted performance. The figure shows the location of a 5G mmWave radio, which was
mounted on a light pole (PCI 65). The figure also shows the measured RSRP for the 5G mmWave P
cell at three locations, all relatively close to the site. At Location #1 (RSRP = -72 dBm), we turned
and faced the serving cell site. At Location #2 (RSRP = -105.6 dBm) we were facing away from the
cell site, so there was a body between the phone and the radio, plus the backside of the phone faced
away from the site. At Location #3 (RSRP = -80.6 dBm), which was further away from the site
than Location #2, the RF conditions improved because we turned and faced the cell site. If memory
serves us correctly, we turned to face the site because we had noticed the uplink throughput was quite
low and we wanted to see if performance improved if we turned to face the site.
Figure 28. P Cell RSRP and the Impact of Body Blockage and Phone Orientation

3.3 Verizon 5G mmWave Uplink Performance – Sample Results
Recently, we were in Memphis, TN where we inadvertently came across two Verizon 5G mmWave
sites outside of our hotel. Since we observed 4CC uplink at the site, not to mention 8CC downlink,
we captured two short logs from our test vehicle which was located perhaps 200 feet directly in front
of the pole-mounted [Samsung] radio. It was a 3-sector site and we only tested from one location.
Figure 29 includes a time series plot and the average uplink throughput during an uplink Verizon was dedicating
throughput test using the same Umetrix Data Server that we used for our Arizona tests. The PUSCH ~23% more slots to the
throughput on each carrier was largely equivalent with an average total throughput of 644 Mbps. uplink direction, which
partially explains why the
Based on our review of the log file it appears that Verizon was dedicating ~23% more slots to the
uplink throughput was
uplink direction, which partially explains why the uplink throughput was much higher in Memphis much higher in Memphis
than it was in Glendale. than it was in Glendale.
Given the results in Arizona, we also did a short sensitivity study to determine the impact of the
smartphone’s orientation relative to the serving cell site on the uplink throughput. For this study, we
simply held the smartphone outside of the vehicle, starting with the back of the phone facing the cell
site, and then rotating the phone 90° every 30 seconds. For this study there wasn’t any body blockage
although the hand holding the phone still had an influence on performance.
Figure 29. PUSCH Throughput – Total and by Component Carrier (Verizon)

700
644
Total
600
500
400
300
P Cell
200
169 153 159 163
100 S2 Cell S3 Cell
S1 Cell
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 P Cell S1 Cell S2 Cell S3 Cell Total
Time (sec) Average PUSCH Throughput (Mbps)

Figure 30 provides a time series plot of the P Cell RSRP during this test and Figure 31 provides
the average values at each orientation. There was a 1 to 6 dB drop in the RSRP, depending on the
orientation of the smartphone.
Figure 30. RSRP Time Series - by Component Carrier (Sensitivity Study)

RSRP
RSRP (dBm)
(dBm)
−70
−70
Backside Facing
Backside Facing Rightside Facing
Rightside Facing Frontside Facing
Frontside Facing Leftside Facing
Leftside Facing
−75
−75 PP Cell
Cell
−80
−80
S1
S1 Cell
Cell
−85
−85
S2
S2 Cell
Cell
−90
−90
S3
S3 Cell
Cell
−95
−95
00 10
10 20
20 30
30 40
40 50
50 60
60 70
70 80
80 90
90 100
100 110
110 120
120 130
130
Time
Time (sec)
(sec)
Figure 31. Average RSRP - by Component Carrier (Sensitivity Study)

Time #1
Time #1 Time #2
Time #2 Time #3
Time #3 Time #4
Time #4
PP Cell
Cell S1
S1 Cell
Cell S2
S2 Cell
Cell S3
S3 Cell
Cell PP Cell
Cell S1
S1 Cell
Cell S2
S2 Cell
Cell S3
S3 Cell
Cell PP Cell
Cell S1
S1 Cell
Cell S2
S2 Cell
Cell S3
S3 Cell
Cell PP Cell
Cell S1
S1 Cell
Cell S2
S2 Cell
Cell S3
S3 Cell
Cell
−83
−83 −73
−73
−76
−76 −74
−89 −88 −74
−89 −88 −88
−88 −85 −85
−85 −86
−84 −84 −85 −85 −86
−84 −84 −85
−85 −85
−85 −85 −86
−86
RSRP
RSRP (dBm)
(dBm)

The drop in RSRP had a subsequent impact on the uplink throughput, as shown in Figure 32 and
Figure 33. The throughput dropped by up to 40% (right-side facing) compared with the highest
measured throughput (left-side facing). There was also greater variability in the throughput, both
over time and between carriers, for the two orientations with the lowest uplink throughput.
Figure 32. PUSCH Throughput Time Series - by Component Carrier (Sensitivity Study)
175
Backside Facing Rightside Facing Frontside Facing Leftside Facing P Cell
150
175
125 Backside Facing Rightside Facing Frontside Facing Leftside Facing P Cell
150 S3 Cell S1 Cell
100
125
75 S3 Cell S1 Cell
100
50
75
25 S2 Cell
50
0
25 S2 Cell
0 10 20 30 40 50 60 70 80 90 100 110 120 130
0 Time (sec)
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (sec)
591
552
Figure 33. Average PUSCH Throughput - by Component Carrier (Sensitivity Study)
591
552
379
350
379
350
146 140 135 132 157 148 143 143

99 111 99
89 84 79 88 82
146 140 135 132 157 148 143 143
111 99
P Cell S1 Cell S2 Cell S3 Cell Total P99
Cell S189Cell S2
84Cell S3
79Cell Total P Cell S1 Cell 88
S2 Cell 82
S3 Cell Total P Cell S1 Cell S2 Cell S3 Cell Total
Time #2 Time #2 Time #3 Time #4
P Cell S1 Cell S2 Cell S3 Cell Total P Cell S1 Cell S2 Cell S3 Cell TotalMbpsP Cell S1 Cell S2 Cell S3 Cell Total P Cell S1 Cell S2 Cell S3 Cell Total
Time #2 Time #2 Time #3 Time #4
Mbps

3.4 4CC Uplink and the Impact on Current Drain
We did tests from 5 different locations where we measured the impact of uplink mmWave on the The Galaxy S23 delivered
battery current drain. At two of the locations, some of the legacy smartphones struggled to remain the best current efficiency
attached to the mmWave signal so we are only showing results from three of the locations – all rela- and once again the higher
the throughput, the higher
tively ideal from our perspective. Although the results varied by location – largely due to differences
the current efficiency.
in the measured uplink throughput – the current efficiency results were directionally consistent.
The Galaxy S23 delivered the best current efficiency and once again the higher the throughput, the
higher the current efficiency.
We used the same methodology for measuring current efficiency – the measured data speed rela-
tive to the battery current drain required to achieve that data speed (Mbps/mA) – that we’ve used
on numerous occasions over the years. We first characterized the RF conditions with XCAL and
confirmed the presence of the desired RF transmission – in this case the presence of 4CC uplink.
We then unplugged XCAL-Solo/XCAL5 and relied on Umetrix Data to both generate the data
transmission and to measure the throughput in one second time increments. Separately, we used an
Android application to display the smartphone battery’s current consumption and we used another
Android application to record the phone’s display for subsequent playback and analysis.
Lastly, we took into consideration differences in screen size and display brightness by first measuring
the battery current consumption with the display turned on and the smartphone in airplane mode.
We subtracted this current consumption from the current consumption during the uplink data
transfers to achieve an adjusted current consumption, which primarily reflects the current associated
with the data transfer and removes the impact of the phone’s display, which can have a meaningful
impact on overall battery life.
Figure 34 provides a time series plot of the battery current consumption for each smartphone in
airplane mode as well as the average values over the 90-second test. Readers should read too much
into the differences in these values since they could largely reflect differences in the display bright-
ness of each smartphone.
Figure 34. Battery Current Drain in Airplane Mode

Current (mA)
500
334
Galaxy S21 FE
400
263
235
300
Galaxy S22 Ultra 204
200
Galaxy S20 Plus

Galaxy S23
100
0
0 10 20 30 40 50 60 70 80 90 Galaxy Galaxy Galaxy Galaxy
S23 S22 Ultra S21 FE S20 Plus
Time (sec) Average Current (mA)

We did the first test after a late evening Sushi dinner at a location within the Westgate Entertainment
District where we had an excellent view of the serving 5G mmWave radio. We only tested three of
the smartphones, although we repeated the test for two of the smartphones. Figure 35 provides the
uplink application throughput from these tests. We can’t explain the large difference in the uplink
throughput between the two Galaxy S23 tests, but as we mentioned earlier, we felt the mmWave
uplink performance was temperamental. We also note that while we are showing application layer
throughput, we assume it was only generated by 5G mmWave since from our other tests we never
observed LTE throughput when there was 5G mmWave throughput – not always a good thing when
the mmWave throughput was marginal.
Figure 35. Uplink Application Layer Throughput after Dark – by smartphone

Mbps
350
224
300 Galaxy S23 II
250
200
130
Galaxy S23
150
100 Galaxy S21 FE II

Galaxy S20 Plus 56
46
50 31
0 Galaxy S21 FE
0 10 20 30 40 50 60 70 80 90 100 Galaxy Galaxy Galaxy Galaxy Galaxy
S23 S21 FE S21 FE II S23 II S20 Plus
Time (sec) Average (Mbps)

Figure 36 provides a time series plot of the current drain during this 90-second test and Figure The current efficiency for the
37 provides the average adjusted battery current drain in which we subtracted the values shown in Galaxy S23 with 4CC uplink
Figure 34 from the average values from Figure 36. Although the average adjusted current drain for was nearly 10x higher than
the Galaxy S20 Plus, which
the Galaxy S23 and Galaxy S21 FE smartphones increased in the second tests along with the higher
was limited to 1CC uplink.
uplink throughput, the current efficiency still increased or at least remained unchanged due to the
corresponding increase in uplink data speeds. In this test (Figure 38), the current efficiency for the
Galaxy S23 with 4CC uplink was nearly 10x higher than the Galaxy S20 Plus, which was limited to
1CC uplink. The use of different 5G modems was also a likely contributor to the differences in the
current efficiency that we documented.
Figure 36. Battery Current Drain During Uplink Data Transfer after Dark Time Series – by smartphone
mA
2500
Galaxy S20 Plus
Galaxy S21 2111
2000
1500
Galaxy S21, II
Galaxy S23
1000
500
Galaxy S23, II
0
0 10 20 30 40 50 60 70 80 90
Time (sec)
Figure 37. Average Adjusted Battery Current Drain During Uplink Data Transfer after Dark – by
smartphone
1290
917
629 638
490
Galaxy S23 Galaxy S21 FE Galaxy S21 FE II Galaxy S23 II Galaxy S20 Plus
Average Adjusted Current (mA)

0.35
0.21

Figure 38. 5G mmWave Uplink Data Transfer Current Efficiency after Dark – by smartphone
0.35
0.21
0.06 0.06
0.04
Current Efficiency (Mbps/mA)
We did the next set of tests after a Tex-Mex lunch while sitting underneath some large umbrellas
which shaded us from the sun. Figure 39 shows a picture of the serving 5G mmWave cell site from
this test location.
Figure 39. View of the 5G mmWave Radio at Lunch

Figure 40 provides a time series plot of the uplink throughput and Figure 41 shows the average
values during the 4-minute tests. Although we know the Galaxy S22 Ultra supported 4CC uplink,
it didn’t appear that the phone used it during this test. We assume the Galaxy S23 smartphone used
4CC uplink during its test, although it is nearly possible to achieve the measured average data speeds
with a single component carrier of mmWave.
Figure 40. Uplink Application Layer Throughput after Lunch Time Series – by smartphone
Mbps
250
Mbps
250
200 Galaxy S23
200 Galaxy S23
150
150
100
100 Galaxy S22 Ultra Galaxy S21 FE
50 Galaxy S22 Ultra Galaxy S21 FE
50
0 Galaxy S20 Plus

0 0 20 40 60 80 100 120 140 160 180 Galaxy 200
S20 Plus 220 240
0 20 40 60 80 100 Time120(sec) 140 160 180 200 220 240
Time (sec) Source: Signals Research Group
Figure 41. Average Uplink Application Layer Throughput after Lunch – by smartphone
163.1
163.1
P Cell RSRP = -80 dBm
SP Cell RSRP = -80
-87 dBm
to 88 dBm
S Cell RSRP = -87 to 88 dBm
55.2
55.2
23.3
17.0 23.3
17.0
Galaxy S23 Galaxy S22 Ultra Galaxy S21 FE Galaxy S20 Plus
Average (Mbps)
Average (Mbps)

Figure 42 shows the average current drain of each smartphone after subtracting the impact of The Galaxy S23 had 2.6x
each phone’s display and Figure 43 shows the subsequent current efficiency. In this test, the Galaxy higher current efficiency than
S23 had 2.6x higher current efficiency than the Galaxy S22, largely because the S22 seemingly the Galaxy S22, and 12.2x
higher current efficiency
didn’t use 4CC. The Galaxy S23 also had 12.2x higher current efficiency than the Galaxy S20 Plus.
than the Galaxy S20 Plus.
We don’t have results for the Galaxy S21 FE because the phone hit its 5G thermal limit during the
test and shutdown the use of 5G mmWave. As shown in the next chapter, the thermal issue was
primarily due to ambient heat and the partial glare of the sun. When analyzing the Umetrix Data
log file, we noticed the starting temperature of the S21 smartphone’s battery was actually close to
its thermal limit when we started the test even though we hadn’t used the phone prior to this test.
Blame global warming.
Figure 42. Average Adjusted Battery Current Drain During Uplink Data Transfer after Lunch – by
smartphone
1016
1016
579
502
579
502
Galaxy S23 Galaxy S22 Ultra Galaxy S20 Plus

0.28
Figure 43. 5G mmWave Uplink Data Transfer Current Efficiency after Lunch – by smartphone
0.28
0.11
0.11
0.02

Efficiency (Mbps/mA) 0.02

Efficiency (Mbps/mA)

We did this last test while sitting on a park bench, literally a stone’s throw from the 5G mmWave
radio. We also used this site to test at greater distances but since not all smartphones could remain
locked on 5G mmWave, we decided to not publish the results. Figure 44 shows the uplink throughput
for each smartphone during this test. Worth noting, in this test the S22 Ultra used 4CC. In fact, its
uplink throughput was higher than the throughput for the Galaxy S23.
Figure 44. Uplink Application Layer Throughput on Bench – by smartphone

Mbps
400
Galaxy S22 Ultra 285.8
244.9
300
Galaxy S23
200
Galaxy S21 FE
100 72.1
41.4
Galaxy
0 S20 Plus
0 20 40 60 80 100 120 140 160 180 200 220 240 Galaxy Galaxy Galaxy Galaxy
S23 S22 Ultra S21 FE S20 Plus
Time (sec)
Average (Mbps)
Figure 45 provides the average current drain during these tests after adjusting for the impact of
the display. Finally, Figure 46 shows the current efficiency for each smartphone. Although the S22
Ultra had higher throughput in this test, its current efficiency was lower than the Galaxy S23 due to
the S23 having close to 35% lower current.
Figure 45. Average Adjusted Battery Current Drain During Uplink Data Transfer on Bench – by
smartphone
948 972
849
614
0.40
0.30

Figure 46. 5G mmWave Uplink Data Transfer Current Efficiency on Bench – by smartphone
0.40
0.30
0.07
0.05
Current Efficiency (Mbps/mA)

3.5 5G mmWave Data Transfers and the Impact on Battery
Temperature
Since XCAL-Solo captured Android OS reported metrics, like battery current drain and battery
temperature, and since the 5G thermal issue is a hot topic, we are including some results which show
the impact of 5G usage [and sunlight] on battery temperature.
After our battery current testing following lunch, we walked around the entertainment district We used ice cubes to
while doing some more uplink testing. During this testing we noticed the Galaxy S23 smartphone’s periodically cool the Galaxy
battery temperature was quickly getting close to the 5G thermal limit, or nearly 42° Celsius. To S23 smartphone while
measuring the impact on 5G
continue testing, and out of curiosity, we grabbed a cup of ice from the restaurant and used the ice
mmWave performance, not
cubes to periodically cool the backside of the smartphone while making sure the water droplets to mention the reported
didn’t work their way into the USB port on the phone. Warning! Do not try this at home! We’re battery temperature.
Professionals!
Figure 47. Taking a Refreshing Ice Bath
Figure 48 shows the results from this study. The figure plots the LTE and 5G mmWave throughput,
including the total 5G and each component carrier, along the primary Y axis. The secondary Y axis
plots the battery temperature. At the start of the test, the battery temperature was 28.6° C and by
400 seconds into the test the temperature had increased to 40.2° C, or 104.4° F, which was close to
the 5G thermal trigger, which generally occurs around 42° C. Around 340 seconds into the test the
smartphone stopped using 5G mmWave with all the uplink data traffic reverting to LTE, but this
situation was due to poor 5G mmWave RF conditions versus thermal.
Thanks to a well-placed ice cube that massaged the backside of the smartphone, the temperature Thanks to a well-placed ice
quickly dropped to 26.4° C, or 79.5° F. The 5G mmWave throughput resumed during this time, but cube that massaged the
we believe it was more coincidental and due to an improvement in the 5G mmWave RF conditions. backside of the smartphone,
the temperature quickly
Since this low temperature was well below the normal operating temperature of the smartphone and
dropped to 26.4° C.
since the sun was shining on the phone’s display, the temperature quickly jumped and then gradually
started increasing due to a combination of sunlight and 5G mmWave usage.

At 600 seconds, the Galaxy S23 stopped using 4CC uplink and reverted to 1CC uplink. Given
the temperature of the smartphone was 41.4° C and the RF conditions were excellent at the time, it
is safe to assume the smartphone was using thermal mitigation techniques. After another ice cube
treatment, the Galaxy S23 resumed its use of 4CC uplink for the remainder of the test.
Figure 48. 5G mmWave Uplink Data Transfers and the “Ice Cube” Effect
Mbps Battery Temp (Celsius)
400 45
5G mmWave Total
Temp
The “Ice Cube” Effect
300
The “Ice Cube” Effect 30
200
S2 Poor 5G mmWave
RF (drop) 15
PUSCH S1
100
PUSCH
LTE
PUSCH S3 PUSCH
P PUSCH
0 0
0 50 100 150 200 250 300 350 400 450 500 550 600 650
1CC due to thermal mitigation
Time (sec)
Mbps Battery Temp (Celsius)

100 45
80
4CC returns due to 30
"ice cube"
60
40
15
20
Secondary carriers dropped due to
temperature (RF conditions were excellent)
0 0
500 510 520 530 540 550 560 570 580 590 600 610 620 630 640 650 660
Time (sec)

After the sun had set, we did some additional walk tests in the parking lot. In these tests the rate
of increase in the battery temperature was far more tolerable even though the outside air temperature
was still near 80° F. Figure 49 shows results from a downlink walk test and Figure 50 shows results
from an uplink walk test. During the downlink walk test, the Galaxy S23 downloaded 170 GB of
data over 5G in just under 18 minutes, including a brief period over Band n77. During this time, the
battery temperature increased 6.7° C with no assistance from an ice cube. During the uplink walk
test, the same smartphone transferred 9.4 GB of data over 5G mmWave and Band n77 over a period
of nearly 18 minutes. Once again, the temperature increase was relatively modest, or just under 6° C.
We note the backlight was enabled during both tests.
The results from these two tests help demonstrate that while thermal issues do exist with 5G, the
problem isn’t specific to the 5G modem and processor. Instead, we attribute thermal-related issues
to a combination of the 5G modem/processor, the phone’s backlight display, body temperature, the
ambient temperature, and especially sunlight.
Figure 49. 5G mmWave Downlink Data Transfers

Total 5G throughput (Mbps) Battery Temperature (Celsius)
3500 42
6.7° Celcius increase in Temperature in 17.9 minutes
170 GB of Transferred Data Battery Temperature
3000 40
2500
38
5G Total Throughput
2000
36
1500
34
1000
Band n77 32
500
0 30
0 100 200 300 400 500 600 700 800 900 1000 1100
Time (sec)
Figure 50. 5G mmWave Uplink Data Transfers

Total Data Transferred (MB) Celsius
10,000 40
Battery Temperature (Celcius)
8,000 38
Total Data Transferred

6,000 36
4,000 34
2,000 9.4 GB Transferred 32

5.9 ° Increase in Temperature
~18 minutes of continuous uplink data transfers
0 30
0 100 200 300 400 500 600 700 800 900 1000 1100
Time (sec)

3.6 5G mmWave Downlink Performance
Our primary purpose for this benchmark study was to characterize 5G mmWave uplink perfor-
mance with 4CC. We did, however, take some time to do some downlink testing – both at the State
Farm stadium (next section) and around the Westgate Entertainment District.
Figure 51 shows the route we used during the test. Each color represents a different 5G mmWave
radio (PCI) although black signifies locations where the smartphone used Band n77. Although it
isn’t evident in this figure, there were several trees obstructing our view of the closest 5G mmWave
radio located at the transition from white to cyan circles, and which resulted in the smartphone using
Band n77 along the longer stretch of black circles.
Figure 51. 5G mmWave Downlink Walk Test
Band n77
PCI 26
PCI 57
PCI 65
PCI 66
PCI 70
PCI 75
PCI 80
PCI 109

We created a few scatter plots to show the downlink performance. Figure 52 plots the 5G mmWave
P cell throughput as a function of the P cell SINR and Figure 53 plots the total 5G mmWave
throughput as a function of the P cell SINR. In both figures, there is a good correlation between
SINR and throughput, as should be expected.
Figure 52. P Cell 5G mmWave PDSCH Throughput Versus P Cell SINR

P Cell Throughput (Mbps)
600
P Cell Throughput (Mbps)
600
500
500
400
400
300
300
200
200
100
100
0
0 −10 −5 0 5 10 15 20 25 30
−10 −5 0 5 10
SINR (dB) 15 20 25 30
SINR (dB) Source: Signals Research Group
Figure 53. Total 5G mmWave PDSCH Throughput Versus P Cell SINR

Total 5G mmWave Throughput (Mbps)
3500
3500
3000
3000
2500
2500
2000
2000
1500
1500
1000
1000
500
500
0
0 −10 −5 0 5 10 15 20 25 30
−10 −5 0 5 10 15 20 25 30
SINR (dB)
SINR (dB)

There was poor correlation between RSRP and the distance to the cell site (Figure 54), as well
as poor correlation between the downlink throughput and the distance to the cell site (Figure 55).
We attribute this outcome to the changing orientation of the smartphone relative to the serving cell
site as well as body blockage. Case in point – we highlighted several data points in Figure 54 which
show a steep drop in RSRP with very little change in distance. This outcome occurred when we went
from facing the serving cell site to facing the opposite direction before walking away from the site.
The subsequent steep drop in throughput occurred while walking away from the site. At roughly 40
meters from the site and 65 meters from the site we turned back to face the cell site and there was a
subsequent jump in RSRP.
Figure 54. P Cell RSRP Versus Cell Site Distance – PCI 65

Facing cell
RSRP (dBm) then turning away
from cell
−70
Facing cell
RSRP (dBm) then turning away Turned back
from cell to face cell
−70
−80
Turned back
to face cell
−80
−90
−90
−100
−100
−110
−110
−120
0 20 40 60 80 100 120 140 160 180 200
−120 Distance (m)

0 20 40 60 80 100 120 140 160 Source: Signals
180 Research Group200
Distance (m)
Figure
2500 55. Total 5G mmWave PDSCH Throughput Versus Cell Site Distance – PCI 65
2500
2000
2000
1500
1500
1000
1000
500
5000
0 20 40 60 80 100 120 140 160 180 200 220
0 Distance (m)
0 20 40 60 80 100 120 140 160 180 200 220
Distance (m)

Figure 56 provides a geo plot of the total throughput during the walk test. The white star shows
the location of PCI 65. The three white squares show locations where we turned to face this 5G
mmWave radio. In each instance, there is a noticeable change in colors, reflecting the sharp increase
in the downlink throughput.
Figure 56. Total 5G mmWave PDSCH Throughput Geo Plot
(Mbps)
No mmWave
throughput
= <250
= < 500
= < 750
= < 1000
= < 1500
= < 2000
= < 2500
= > 2500

3.7 State Farm Stadium Performance Results
We did these tests on our first day in Arizona and just prior to the start of the US-Mexico soccer
match. When we did these tests, the stadium was relatively unpopulated, so we were able to move
quite easily throughout the stadium. Plus, the light attendance at the time suggested network traffic,
especially on 5G mmWave, was very modest. We had a temporary software issue which prevented us
from doing walk tests, so we reverted to doing several stationary tests to show 5G mmWave perfor-
mance. Further, since the “before” network configuration was being used, we rarely encountered
4CC uplink. We only observed it at three locations and the 5G mmWave uplink throughput only
exceeded 100 Mbps at one location.
Despite the lower than anticipated throughput, we did encounter a large number of unique 5G
mmWave PCI values (radios), as reflected in Figure 57. There were also a few locations where the
smartphone reverted to Band n77 or Band n2. Our testing included close to 10 additional locations
but for some reason the files got corrupted, so we don’t have any data to show for these locations.
Figure 57. Stadium Map PCI Values
Band n77 247

Band n2 250
Band n2/n77 252
359 253
241 228/242/254
205 245
207 246
208 248
210 249
211 237/241/253/254
232/248 290 (seat for
game)

Figure 58 shows the average 5G mmWave downlink throughput at each test location. We note The current software
there wasn’t any data traffic over the LTE connection when the smartphone was using 5G mmWave, release from Nokia doesn’t
even if the data speeds were modest. LTE did contribute to the total throughput with very low 5G scale to support both 600
MHz and a large number
mmWave data speeds. We also encountered some 5G mmWave radios that were limited to 4CC in
of connected devices.
the downlink even though AT&T had access to 600 MHz of spectrum. This situation also occurred
at our seat during the game. It is our understanding that this limitation is a current phenomenon
since the current software release from Nokia doesn’t scale to support both 600 MHz and a large
number of connected devices. Hence, those mmWave radios covering a larger geographic area within
the stadium were more likely to be limited to 400 MHz of spectrum so that they could support a
greater number of connected smartphones.
Figure 58. Stadium Map 5G Downlink Throughput
= < 100 Mbps

= < 500 Mbps
= < 1000 Mbps
= < 1500 Mbps
= < 2000 Mbps
= > 2000 Mbps

Figure 59 shows the average SINR at each test location. Figure 60 shows the average 5G
throughput at each test location. Finally, Figure 61 shows the average RSRP at each test location.
The low RSRP at many locations, combined with the initial network configuration, helps explain
why the Galaxy S23 rarely used more than 1CC for uplink data transmissions within the stadium.
The downlink buffer requirement was also a big contributing factor. With PCI 210 (RSRP = -73.2
dBm) the phone used 4CC for its uplink data transfer, but the uplink throughput was still less
than 200 Mbps. At many locations, the uplink throughput was, at best, in the low double digits
when using mmWave. These speeds can also be delivered over LTE, especially with uplink carrier
aggregation. However, mmWave does provide additional capacity that LTE can’t deliver with its
limited amount of spectrum.
Figure 59. 5G P Cell SINR by PCI

21
18 17
16.9
15.2 14.9
13.7
11.5 11.9
9.8 10.1 10.8
8.4 21 9.3
7 6.1 18 17
3.5 16.9
15.2 14.9 1.8
13.7
0 11.9
11.5 10.8
9.8 -0.3 10.1 9.3
8.4 -1.6
-4 7 6.1 -3.9
-7.8 3.5
1.8 -9.3
0 -9.8
-0.3
-1.6
228/242/254
Seat for Game

232/248
247
250
252
253
EndZone (n77)
EndZone (n77)
(n2/n77)
245
246
207
249
250
252
359
241
247
n77
205
208
210
211
n2
n2
-4 -3.9
-7.8
-9.8 -9.3
SINR (dB)
228/242/254
232/248
247
247
Seat for Game

250
252
253
EndZone (n77)
EndZone (n77)
(n2/n77)
245
246
207
249
250
252
n77
359
241
205
208
210
211
n2
n2
2448
SINR (dB) 2068

Figure 60. 5G Total Throughput by PCI
1746
2448
1545
1401 1384 1353
1303 1306 1250 2068
1140
1061 1039
948 1746
819
1545 698
1401 1384 1353 537
521 1303 1306
418 1250
280 1140
1061 1039
948 131 80
819 53 42.2 13.7 25.2
698
228/242/254
232/248
247
247
Seat for Game

250
252
253
EndZone (n77)
EndZone (n77)
(n2/n77)
245
246
207
249
250
252
n77
359
241
205
208
210
211
n2
n2
521 537
418
280
131 80
53 42.2 13.7 25.2
Mbps
228/242/254
Seat for Game

232/248
247
250
252
253
EndZone (n77)
EndZone (n77)
(n2/n77)
245
246
207
249
250
252
n77
359
241
247
205
208
210
211
n2
n2

Mbps

Figure 61. 5G P Cell RSRP by PCI
Seat for Game

EndZone (n77)
EndZone (n77)
228/242/254
(n2/n77)
232/248
208
246
249
250
250
205
n77
207
245
359
252
252
247
253
247
210
241
211
n2
n2
-73.2
-82.1 -79.8
-85 -86 -84 -83.1
-90 -90 -88 -90 -89 -88 -87.7
-91 -94 -91 -92 -93.3
-97 -95.5 -96.4
-99 -99
-105 -106
RSRP (dBm)

4.0 Test Methodology
Consistent with virtually all our 5G benchmark studies, we collaborated with Accuver Americas and We collaborated with
Spirent Communications – two trusted partners that we have worked with for nearly 15 years. We Accuver Americas and
used XCAL-Solo to collect the chipset diagnostic messages and we used the XCAP post-processing Spirent Communications
– two trusted partners
tool to analyze the results. Spirent Communications provided its Umetrix Data platform which
that we have worked with
we used to generate high bandwidth data transfers and to capture the application layer throughput for nearly 15 years.
during our current efficiency tests.
We used Umetrix Data to generate the uplink and downlink data traffic. For both directions,
we used a full buffer HTTP data transfer. Each test cycle lasted 4 minutes and the cycles repeated
continuously throughout each walk test. For the current efficiency tests, we used a single 4 minute
test and occasionally a 90 second test.
Figure 62. Umetrix Data Platform
Source: Spirent Communications
For the current analysis tests, we used a proven, albeit cumbersome, methodology that we have used
numerous times in the past. We did these tests while stationary and with one phone at a time. While
conducting the uplink data test with Umetrix, we used a third-party Android application to display
the instantaneous current drain of the smartphone to determine how much current the smartphone
required during the test. Since it would be logistically challenging to capture these values in real
time, we used a video screen capture application to record the phone’s display. Following the testing,
we used video editing software to play back each video while we manually wrote down each current
value. We also recorded the current while the smartphone was in airplane mode with the backlight

display on so that we had a baseline current drain which would allow us to calculate a net current
drain, or just the incremental current required to support the data transfer.
Lastly, we used the results stored on the Umetrix data server to determine the application layer Once we had the average
throughput during these tests. Umetrix provides application layer throughput in one-second time throughput and the average
increments along with other useful information. Once we had the average throughput and the average net current, it was a simple
calculation to arrive at
net current, it was a simple calculation to arrive at the current efficiency (current efficiency = average
the current efficiency.
throughput/average net current). We note Umetrix only captures application layer throughput
but since all the data transferred over 5G mmWave (no LTE), we know all the application layer
throughput was due to 5G mmWave.
XCAL-Solo is a handheld unit that connects to the smartphone via the USB port, thus making it
ideal for walk tests and testing inside an NFL stadium which bans personal computers. XCAL-Solo
captures the chipset diagnostic messages that are reported by the 5G modem, thus providing a
wealth of information about the LTE and 5G RF conditions as well as how the network is sched-
uling the smartphone with LTE and 5G network resources for each frequency band being used by
the smartphone. This information is reported on a per slot basis although for our analysis we used
XCAP to bin the data into one-second time increments. XCAL-Solo also logs information reported
by the Android operating system so we used this feature to record the battery temperature of the
smartphone.
When testing outdoors, we used the GPS capabilities of the smartphone to track our location. For
the indoor tests within State Farm stadium, we used the inbuilding mapping features of XCAL-Solo
to depict where within the stadium we were testing. In summary, we used a preloaded JPG map of
the stadium to mark our location at each test point while with indoor mobility tests it is possible to
trace out the walk path and then overlay the chipset measurement reports over the path between
each way point.
Figure 63 shows a picture of the handheld XCAL-Solo unit that we used for our walk tests. We
kept the Solo unit detached from the smartphone during our tests so that it wouldn’t interfere with
the 5G mmWave signals.
Figure 63. XCAL-Solo
Source: Accuver Americas

5.0 Final Thoughts
Yes, we know it has been a while since our last report, but we haven’t forgotten our loyal readers.
We still have much planned for the coming months, including uplink 5G carrier aggregation, more
MU-MIMO, and Open RAN. Until next time, be on the lookout for the next Signals Ahead….
Michael Thelander
Michael Thelander is the President and Founder of Signals Research Group (SRG), a US-based
research consultancy that offers thought-leading field research and consulting services on the
wireless telecommunications industry.
Its flagship research product is a research product entitled Signals Ahead, which has attracted
a strong following across the entire wireless ecosystem with corporate subscribers on five
continents. SRG’s Signals Ahead research product and its consulting services are technology-
focused with a strong emphasis on next-generation networks and performance benchmarking.
In his current endeavor, Mr. Thelander is the lead analyst for Signals Ahead and he guides
a team of industry experts that provide consulting services for the wireless industry, including
some of the largest mobile operators, the top equipment OEMs, trade associations, and finan-
cial institutions. He has also served as a member of an industry advisory board for one of the
world’s largest wireless infrastructure suppliers.
Mr. Thelander earned a Masters of Science in Solid State Physics from North Carolina
State University and a Masters of Business Administration from the University of Chicago,
Graduate School of Business.

ON THE HORIZON: POTENTIAL SIGNALS AHEAD/SIGNALS FLASH! TOPICS
We have identified a list of pending research topics that we are currently considering or presently working on
completing. The topics at the top of the list are definitive with many of them already in the works. The topics toward
the bottom of the page are a bit more speculative. Obviously, this list is subject to change based on various factors and
market trends. As always, we welcome suggestions from our readers.
Thematic Reports
➤ Mobile Edge Computing and the impact of data caching at the cell edge
Benchmark Studies
➤ 5G NR mmWave Fixed Wireless Access with IAB
➤ UL-MU-MIMO
➤ 4CC uplink mmWave benchmark study
➤ Open RAN network performance benchmark study 1 – Dish Network Revisit
➤ Open RAN network performance benchmark study 3 – Scheduling Efficiency
➤ FR1 + FR2 NR-DC network performance benchmark study
➤ MU-MIMO benchmark study, part II (FR1)
➤ SRS-based beamforming benchmark study
➤ 5G mmWave device/chipset lab-based benchmark study
➤ DSS Update benchmark study

Signals Ahead Subscription
The Signals Ahead newsletter is available on a subscription basis. We offer three distinct packages that have been
tailored to address the needs of our corporate users. The Group License includes up to five users from the same company.
The Global License is the most attractive package for companies that have several readers since it is offered to an unlimited
number of employees from the same organization. Finally, the Platinum package includes the Global License, plus up to five
hours of analyst time. Other packages are available.
Corporate Rates (12 issues)

❒ Group License ($3,995)
❒ Global License (Price Available upon Request)
❒ Platinum (Price Available upon Request)
❒ Gold Pass (Price Available upon Request)
Payment Terms
❒ American Express ❒ Visa ❒ MasterCard Credit Card # Exp Date //
❒ Check Check Number
❒ Purchase Order PO Number
Name: Title:
Affiliation: Phone: ( )
Mailing Address:
Mailing Address
Signals Research Group – ATTN: Sales
5300 Painter Creek Green
Independence, MN 55359
Alternatively, you may contact us at (510) 273-2439 or at [email protected] and we will contact you for your
billing information.
Terms and Conditions: Any copying, redistributing, or republishing of this material, including unauthorized
sharing of user accounts, is strictly prohibited without the written consent of SRG.
please note disclaimer: The views expressed in this newsletter reflect those of Signals Research Group and are based on our understanding of past and current events shaping the wireless industry.
This report is provided for informational purposes only and on the condition that it will not form a basis for any investment decision. The information has been obtained from sources believed to be
reliable, but Signals Research Group makes no representation as to the accuracy or completeness of such information. Opinions, estimates, projections or forecasts in this report constitute the current
judgment of the author(s) as of the date of this report. Signals Research Group has no obligation to update, modify or amend this report or to otherwise notify a reader thereof in the event that any matter
stated herein, or any opinion, projection, forecast or estimate set forth herein, changes or subsequently becomes inaccurate.
If you feel our opinions, analysis or interpretations of events are inaccurate, please fell free to contact Signals Research Group. We are always seeking a more accurate understanding of the topics
that influence the wireless industry. Reference in the newsletter to a company that is publicly traded is not a recommendation to buy or sell the shares of such company. Signals Research Group and/or
its affiliates/investors may hold securities positions in the companies discussed in this report and may frequently trade in such positions. Such investment activity may be inconsistent with the analysis
provided in this report. Signals Research Group seeks to do business and may currently be doing business with companies discussed in this report. Readers should be aware that Signals Research Group
56 have a conflict of interest that could affect the objectivity of this report. Additional information and disclosures can be foundMay
might at our10, 2023
website | Signals Ahead, Vol.
at www.signalsresearch.com. This19, Number
report 3
may not be
reproduced, copied, distributed or published without the prior written authorization of Signals Research Group (copyright ©2022, all rights reserved by Signals Research Group).

SA Vol 32 4CC Uplink

Uploaded by

Copyright:

Available Formats

SA Vol 32 4CC Uplink

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SA Vol 32 4CC Uplink

Uploaded by

Copyright:

Available Formats

Redefining Research May 10, 2023

Key Highlights from this Study

3 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

4 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

➤ KPIs analyzed include:

♦ Distribution of uplink throughput by component carrier

♦ Uplink throughput versus PUSCH transmit power

♦ PUSCH transmit power versus RSRP

♦ Impact of smartphone orientation and body blockage on RSRP and throughput

➤ Mapping out the 39GHz performance in State Farm stadium

➤ Comparative results on the Verizon 28 GHz network

5 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

6 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Index of Figures & Tables

7 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

8 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

9 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

10 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

11 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Although heavy 5G mmWave usage has an impact on battery temperature, in

12 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

We believe operators will continue to adjust the ratio of downlink versus

13 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

14 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

3.1 Before and After

Figure 1. 5G mmWave and Band n77 Coverage - Before

Source: Signals Research Group

15 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Source: Signals Research Group

Cumulative Probability Distribution (%)

2.8 2.7 2.8

P Cell S1 Cell S2 Cell S3 Cell Total

16 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Cumulative Probability Distribution (%)

2.8 2.7 2.8

P Cell S1 Cell S2 Cell S3 Cell Total

Figure 6. Distribution of Uplink Throughput – Total and by Component Carrier (After)

17 71.1 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Figure 7. Average and Distribution of RSRP –by Component Carrier (Before)

Figure 8. Average and Distribution of RSRP – by Component Carrier (Before)

18 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Figure 9. Geo Plot of 5G P Cell RSRP (Before)

Source: Signals Research Group

Figure 10. Geo Plot of 5G P Cell RSRP (After)

Source: Signals Research Group

19 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

20 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

21 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

−110 −105 −100 −95 −90 −85 −80 −75 −70

−110 −105 −100 −95 −90 −85 −80 −75 −70

22 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

23 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Source: Signals Research Group

Source: Signals Research Group

24 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Figure 23. Distribution of 5G and Component Carrier Usage

1CC (>1 Mbps)

4CC (>1 Mbps per CC) 7%

25 May 10, 2023 | Signals Ahead, Vol. 19, Number 3

Figure 24. Transmission Type Versus RSRP (After)