Enabling Coexistence Between Multiple Radio Access Technologies

The radio frequency (RF) spectrum must support a growing number of users and applications across various wireless networks, including cellular systems, local-area networks, industrial wireless systems, and IoT deployments. As these technologies increasingly operate in shared frequency bands, intelligent use of spectrum resources becomes essential for efficient wireless coexistence. In our work, we look into how AI can support this goal in current and future cellular networks built on the Open Radio Access Network (O-RAN) architecture.

O-RAN is an open, software-based architecture that enables custom intelligent applications to monitor network conditions and adapt network operation, e.g., its radio resource utilization. In particular, we focus on distributed applications (dApps), which can be deployed directly on RAN nodes and access radio-level data to support real-time network monitoring and control. We use an OpenAirInterface simulation testbed to study how dApps can be used for spectrum management in a coexistence scenario involving cellular and Wi-Fi technologies operating in the same frequency band. The dApp analyzes raw radio signals received at the base station to identify active wireless transmissions and their spectral locations. Then, based on this information, it configures cellular transmissions to use only unoccupied portions of the spectrum, thus supporting interference-aware coexistence among networks.

To provide the spectrum awareness needed for this adaptation, the dApp employs SigDETR, an AI-based module we developed to analyze spectrum occupancy. Using a Transformer-based deep learning architecture, SigDETR processes raw I/Q samples of the sensed radio signals to detect, classify, and spectrally localize wireless transmissions across a wide band in an end-to-end manner. This yields the spectrum occupancy information that drives the dApp’s closed-loop spectrum management decisions.

Further details on SigDETR’s architecture and its spectrum analysis capabilities are presented in our recent paper, published at IEEE DySPAN 2026: A. Vagollari, S. S. Sawant, S. Raghunandan, and W. Gerstacker, “A Transformer Network for End-to-End Detection and Classification of Wideband RF Signals,” 2026 IEEE International Symposium on Spectrum Innovation (DySPAN), Washington, DC, USA, 2026, pp. 1-6, doi: 10.1109/DySPAN69846.2026.11571159.