日本語 
SoftBank, Ericsson Japan, and Qualcomm Technologies ran a field trial in Tokyo on a live commercial 5G SA network. Not a demo network. Not a lab. Actual city traffic. The focus was low latency communication using a mix of 5G and 5G-Advanced features, including L4S. The use case was XR streaming on smart glasses. They picked XR for a reason. Even small delays break the experience.
The trial was about one question. Can current 5G SA networks actually support real-time XR without choking.
The setup was practical. Smart glasses were connected to a smartphone over Wi-Fi. That smartphone connected to a commercial 5G base station in Tokyo. Data moved between the devices and an application server on the internet. Nothing fancy. Just real conditions.
XR content was streamed end to end. This matters because XR needs near instant response. Any delay shows up as stutter, lag, or motion mismatch. That is why XR was chosen to validate low-latency performance.
Also Read: KOSA and MegazoneCloud Aid 13 AI Firms Enter Japan
The outcome was clear. Wireless section latency was reduced by around 90 percent compared to when L4S and other advanced 5G features were not used. The communication stayed continuous and stable. Low latency was not a short spike. It held throughout the test.
They also used network slicing. The advanced 5G and 5G-Advanced features were applied only to the devices in the field trial. That allowed the network to be optimized specifically for XR streaming without affecting other users on the same commercial network.
Several technologies were tested together.
L4S was used to reduce communication delay. It works by using ECN bits to detect early signs of congestion. Transmission rates are adjusted in real time. The idea is simple. Fix the problem before congestion gets bad. This prevents packet buildup and keeps latency low and stable.
Configured Uplink Grant focused on the uplink side. It simplifies how devices send data to the base station. Devices can transmit immediately based on predefined schedules and conditions. No repeated requests. This cuts uplink delay, which is critical for XR where devices constantly send motion and control data.
Scheduler-based rate control was handled at the base station. Radio resources were dynamically controlled to maintain the required throughput. This helped keep XR visuals stable and made interaction responses smoother.
Network slicing played a supporting role. The network was virtually divided so the XR service could get the quality it needed. Low latency and reliability were prioritized for the trial while the rest of the network continued operating normally.
Each company had a defined role.
SoftBank selected the base stations used for the trial. They planned the evaluation items, methods, and studied real communication conditions at 5G stations across Tokyo.
Ericsson provided the 5G and 5G-Advanced technologies, including L4S. They ran preliminary verification and configured the functions on the base stations used in the trial.
Qualcomm Technologies provided the complete hardware and software stack. This comprised of smart glasses using the Snapdragon AR2 Gen 1 platform, a mobile phone operating on the Snapdragon 8 Elite Mobile Platform with the cutting-edge Qualcomm 5G Modem-RF system, and the XR application that was part of the test scenario.
This was not about future promises. It was a live network, real devices, and one of the hardest real-time use cases to support. The latency numbers dropped. The experience stayed stable. That is the takeaway.
English