We align what devices see with what the radio channel does, so next-generation networks can anticipate XR users instead of reacting to them.
XR headsets need high throughput, very low latency, and reliable links. In real rooms, people move, turn their heads, and block signals constantly. SCOPE-XR builds an AI-native, integrated sensing-and-communication framework. It fuses on-device perception (pose, gaze, scene geometry) with radio measurements to predict channel conditions and adapt the network ahead of time.
Extended reality (XR), including augmented, virtual, and mixed reality, will require wireless networks that can respond reliably to user motion, changing environments, and strict delay constraints. Existing Wi-Fi and 5G systems remain largely reactive and are not designed to support immersive, multi-user experiences that depend on stable spatial alignment and low motion-to-photon latency. This project will advance a new class of perception-aware wireless systems that will help future networks better support education, healthcare, manufacturing, and public-safety training. By enabling more robust and adaptive XR connectivity, the research will strengthen U.S. leadership in next-generation (NextG) wireless and immersive technologies, broaden access to advanced digital experiences, and train students at the intersection of wireless networking, machine learning, and XR systems. The project will also produce open software and evaluation assets that will support research, workforce development, and future technology translation.
The project will develop an Artificial Intelligence (AI)-native, integrated sensing and communication (ISAC) framework for next-generation XR networking. The central technical idea is a Radio Radiance Field, a vision-aligned channel abstraction that will combine radio measurements with on-device perception to infer predictive channel priors in dynamic environments. These priors will drive software-defined network control functions for beamforming, scheduling, handover, and edge-assisted adaptation through Open Radio Access Network (O-RAN)-compliant xApps and rApps. The project will also couple XR pose, gaze, and rendering priorities with wireless control to enable closed-loop coordination between the sensing, communication, and application layers. The research team will validate the system in representative collaborative learning and industrial XR scenarios and will quantify communication gains together with latency and compute overhead. Expected outcomes will include deployable O-RAN software modules, benchmarked multimodal datasets and evaluation methods, and design principles for applying AI-native and ISAC-enabled control to XR and other perception-driven NextG wireless systems.
Source: NSF Award #2554332
SCOPE-XR is underway. Milestones will be added as the project progresses.
NSF VINES Track 2 award #2554332 begins.
All five partner organizations meet; communication plan and reporting rhythm set.
Project website, cross-site technical planning, and shared data infrastructure.
Five organizations spanning NextG wireless, XR systems, and large-scale testbeds.
Haijian Sun (PI)
Zhenlin An (Co-PI)
Kyle Johnsen (Co-PI)
Mehmet Kurum (Co-PI)
WenZhan Song (Co-PI)Publications, software, and datasets will be listed here as they are released.
All SCOPE-XR code lives on GitHubSoftware, O-RAN xApps/rApps, and tools from every partner site are hosted under the github.com/SCOPE-XR organization.Peer-reviewed papers from the project.
Paired XR perception and radio-channel datasets.
