5G Upload Boost Fails in Real-World Tests

As social media, live streaming, and remote work push mobile networks to handle more upload traffic than ever, telecom companies have turned to advanced 5G technologies to keep pace. Two features in particular—Uplink MIMO and Uplink Carrier Aggregation—promise to double upload speeds by using different technical approaches. But a new real-world evaluation on a commercial 5G network in the United States delivers a counterintuitive finding: in most scenarios, the newer Uplink MIMO technology actually yields slower upload speeds than the older Carrier Aggregation . This s assumptions about how to optimize networks for the upload-heavy applications that are becoming central to modern mobile use.

The researchers, testing T-Mobile's 5G network in Toledo, Ohio, found that Uplink Carrier Aggregation consistently outperformed Uplink MIMO across urban, suburban, and rural environments. Uplink MIMO is designed to transmit two data streams on a single frequency band, theoretically doubling throughput in strong radio frequency conditions. In contrast, Uplink Carrier Aggregation allows a device to upload simultaneously on two different frequency bands, each with one data stream, also totaling two streams. The study measured physical layer uplink throughput and normalized the data to account for differences in channel bandwidth and network load, revealing that Carrier Aggregation provided higher and more reliable speeds. For example, in urban areas, the 10th percentile throughput for Carrier Aggregation using bands n41 and n25 was 65.8% greater than for Uplink MIMO, as shown in Figure 3d.

To conduct their evaluation, the team performed drive and walk tests along three representative routes: an urban downtown core, a suburban freeway and local road mix, and a rural area with farmland. They used Samsung and Pixel smartphones with modified firmware to enable or disable specific features, ensuring accurate comparisons. Data was collected using software like AirScreen and Network Signal Guru, with continuous upload stress tests to measure throughput. The researchers filtered out non-representative data points, such as those during network handoffs or when secondary carriers were inactive, to maintain consistency. This ology allowed them to assess performance under varied radio frequency conditions, from strong signals in dense urban settings to weaker ones in rural areas.

, Detailed in Figures 3a through 3f, show that the performance gap widens in weaker RF conditions. In rural tests, the 10th percentile throughput advantage for Carrier Aggregation over Uplink MIMO expanded to 135%. This disparity is partly due to Uplink MIMO's reliance on favorable signal strength; when the signal weakens, devices often revert to a single-stream transmit diversity mode for reliability, negating the speed benefits. For instance, in the rural route, Uplink MIMO used this single-stream mode 89.5% of the time. Additionally, Uplink MIMO struggled with higher-order modulation schemes like 256QAM, using it only 5.2% of the time in urban tests compared to 18.6-21.6% for Carrier Aggregation, as seen in Figure 4a, because splitting power between two streams reduces signal quality.

These have significant for network operators and device manufacturers aiming to enhance user experiences for applications like video conferencing, live streaming, and fixed wireless access. The study suggests that prioritizing the adoption of Uplink Carrier Aggregation could offer more consistent upload speeds, especially in areas with variable signal strength. However, Uplink MIMO remains viable in dense urban environments where strong RF conditions allow it to operate efficiently, conserving network capacity by avoiding the use of multiple carriers for a single user. Operators might implement dynamic traffic-steering policies, favoring Uplink MIMO in congested cells to maximize overall sector capacity while reserving Carrier Aggregation for users in weaker RF conditions.

Despite its insights, the study has limitations. It focused on a specific network configuration in one geographic region, and may not generalize to all 5G deployments worldwide. The research did not account for interference from other users, which can affect spectral efficiency, and it evaluated performance without considering real-world application impacts like battery life or live streaming quality. Additionally, the paper notes that most commercial networks and devices do not yet support Uplink Transmit Switching, which combines both features, limiting the current comparison to mutually exclusive options. Future work will explore this combined approach and assess effects on power consumption and specific use cases.