Support Vector Machine Based Spectrum Sensing in Beyond 5G Wireless Networks

Avstract-The primary concern in 5G and beyond wireless technologies revolve around the exponentially growing number of network users and subsequently increasing demand for frequency spectrum utilization. Traditional static spectrum allocation schemes have proven to be highly inefficient in using the available spectrum. In recent years, Dynamic Spectrum Sharing (DSS) solutions based on Cognitive Radio (CR) and Artificial Intelligence (AI)/Machine Learning (ML) have been proposed. CR involves spectrum sensing, spectrum sharing and decision-making paradigms essential for efficient use of radio frequency spectrum. AI/ML techniques, with their autonomous classification, learning, and decision-making capabilities, offer an improved approach. This paper proposes a Support Vector Machine (SVM) based ML technique for spectrum sensing in beyond 5G networks. A theoretical examination of detection and false alarm probabilities is first conducted. It is shown that the SVM algorithm achieves a high probability of detection at 99.66%, ensuring reliable spectrum sensing. Further, it is verified through simulations that the experimental outcomes align closely with the theoretical results concerning both detection and false alarm probabilities.