On March 3, 2025, NTT Corporation and NTT DOCOMO announced the successful demonstration of the effectiveness and feasibility of In-Network Computing (INC), an architectural solution that integrates computing resources with mobile communication devices.
The proposed architecture coordinates slicing and communication quality control aligned with the activation of computing functions on the User Plane Function (UPF) within the mobile core network.
During the demonstration, the GSMA-defined Open Gateway / CAMARA API was utilized to control network routing based on mobile network conditions. Additionally, new technologies were developed and implemented that enable coordinated control between user communication devices and servers positioned near the user.
Among the anticipated 6G use cases are new services such as immersive XR, artificial intelligence / machine learning, and sensing. In the demo, an AI-powered video analysis application was operated within this environment and by linking the computing services with the mobile network data exchange latency and jitter between devices and servers was reduced. The AI-powered application executed in the network was used to process a video transmission; the accuracy rate of the AI model — which is 57% under conventional architectures — improved to 90%, reaching the performance limit of the AI model.
This proof of concept was conducted in collaboration with Nokia Corporation. The results were showcased at Nokia’s booth during Mobile World Congress Barcelona 2025.
More information about the demonstration is available at
https://www.docomo.ne.jp/english/info/media_center/pr/2025/0303_02.html
NTT DOCOMO and DOCOMO Euro-labs are actively promoting the standardization of INC in the 3GPP 6G architecture. INC is a fundamental enabler of innovative services in the 6G era, which will enable to realize compute intensive services with low latency and high throughput requirements (such as immersive XR and AI-powered applications) making efficient use of communication and compute resources.