A Survey of Spectrum Allocation in Cognitive Radio Using Artificial Intelligence

Abstract

AI-driven spectrum allocation in Cognitive Radio (CR) is one of the most prominent paradigms towards enhancing efficiency, adaptability, and scalability of the radio spectrum, addressing limitations of traditional methods like game theory and heuristic algorithms. These conventional approaches struggle with real-time adaptability and efficient spectrum utilization in dynamic environments, which suggest that adopting AI techniques could allow the optimization of decision-making and reduce interference in order to capitalize on the already scarce spectrum, in a period of all-time high demand for resources, enabling a more adaptive spectrum access, improving performance in multi-user environments and aiding in spectrum sensing and prediction, by offering scalable distributed allocation solutions. In this regard, challenges to this new paradigm include computational complexity and security risks. Many AI models rely merely on simulations and proofs of concepts, requiring real-world use case validation, and regulatory compliance with global standards remains a crucial point of discussion. This work provides a survey that underscores AI’s transformative role in CR spectrum management, by outlining the existing traditional methods of spectrum allocation and their shortcomings, and how overcoming existing challenges and integrating emerging AI based techniques represents the forefront to attaining the full potential of spectrum allocation in next-generation wireless networks.