Summary

SRG just completed its 61st 5G benchmark study. For this endeavor we collaborated with Accuver Americas and Keysight Technologies to conduct a benchmark study of FDD massive MIMO. We tested throughout a Verizon 4T4R cell site (Band n2) and then again after the operator had upgraded the cell site to 32T32R with Ericsson infrastructure. We looked at the impact on both downlink and uplink performance.

Highlights of the Report include the following:

Our Thanks. We did this study with the support of Accuver Americas and Keysight Technologies. Accuver provided us with its XCAL5 drive test software and XCAP post-processing software. We used XCAL5 to collect the chipset diagnostic messages on all smartphones and we used XCAP for the post processing of the data and subsequent analysis. We used Umetrix Data from Keysight Technologies to generate the high bandwidth downlink and uplink UDP data transfers. Both companies have been trusted partners with us for nearly two decades. SRG is responsible for all analysis and commentary provided in this report.

The Background. Ericsson had released a few press releases and a whitepaper over the last few months regarding the performance of its massive MIMO FDD radio, including the AIR 3283 in the case of Verizon. We took it upon ourselves to do our own testing and publish the outcome in detail. Verizon identified a cell site in downtown Saint Paul, MN that was going to be upgraded from 4T4R to 32T32R so we were able to easily do a before and after study. Verizon had no other involvement in the study while Ericsson didn’t know what we were up to until just prior to publishing this report. We provided both companies with a short summary presentation just prior to releasing this report. This study was solely funded by our Signals Ahead subscription model.

The Methodology. We used two Motorola razr fold 2026 smartphones. With these smartphones we were able to lock them to Band n2 5G SA to collect meaningful data in our walk tests. We used the same routes when walking the area with the two different radios and then filtered the results so that our analysis focused on the cell sectors (PCI) that were part of the network upgrade. When analyzing the data, we did both geo binning and analysis of various metrics as a function of RSRP (uplink) and SSB-SINR (downlink). We’ll use these phones for some additional testing at this cell site that we have planned for the coming weeks.

The Uplink. Based on geo-binned analysis (5m x 5m grids), we documented close to a triple-digit increase on a percentage basis in resource block efficiency (Mbps per RB). As a function of RSRP, the percentage increase was in the triple digits with edge-of-cell conditions (i.e., low RSRP). Other metrics, including transmit power and uplink MIMO utilization also showed significant improvement with 32T32R. Notably, the Verizon 5G network supported SA and it supported UL-MIMO in an FDD band.

The Downlink. We observed meaningful double-digit gains in total throughput / spectral efficiency on a percentage basis, based on RB normalized throughput with further gains most likely introduced since we did the testing. We plan to revisit downtown Saint Paul to do some retests and to test some new features that have been introduced since we did our testing in late June / early July.

The Implications. Verizon leverages Band n77 (TDD) for the majority of its downlink capacity needs. Band n2 provides additional capacity in both directions, but it is clear its benefit is predominantly in the uplink direction. 32T32R takes it to another level. We also surmise that Band n2 can help extend Band n77 coverage / performance in RF challenging areas when Band n2 is used as the primary cell (FDD-TDD carrier aggregation). We haven’t validated this hypothesis as of yet. Net-Net: Massive MIMO in 5G FDD bands falls into our “no brainer” category of features that operators need to leverage, especially for those high traffic density cell sites.