Decision-Theoretic Distributed Channel Selection for Opportunistic Spectrum Access: Strategies, Challenges and Solutions

• Published in: IEEE Communications surveys and tutorials Vol. 15; no. 4; pp. 1689 - 1713
• Main Authors: Yuhua Xu, Anpalagan, Alagan, Qihui Wu, Liang Shen, Zhan Gao, Jinglong Wang
• Format: Journal Article
• Language: English
• Published: IEEE 01.01.2013
• Subjects: Channel estimation; Cognitive radio; Decision making; distributed channel selection; Game theory; Markov processes; Markovian decision process; multi-armed bandit problem; Opportunistic software systems development; Opportunistic spectrum access; optimal stopping problem; Spread spectrum management
• ISSN: 1553-877X
• DOI: 10.1109/SURV.2013.030713.00189

Opportunistic spectrum access (OSA) has been regarded as the most promising approach to solve the paradox between spectrum scarcity and waste. Intelligent decision making is key to OSA and differentiates it from previous wireless technologies. In this article, a survey of decision-theoretic solutions for channel selection and access strategies for OSA system is presented. We analyze the challenges facing OSA systems globally, which mainly include interactions among multiple users, dynamic spectrum opportunity, tradeoff between sequential sensing cost and expected reward, and tradeoff between exploitation and exploration in the absence of prior statistical information. We provide comprehensive review and comparison of each kind of existing decision-theoretic solution, i.e., game models, Markovian decision process, optimal stopping problem and multi-armed bandit problem. We analyze their strengths and limitations and outline further research for both technical contents and methodologies. In particular, these solutions are critically analyzed in terms of information, cost and convergence speed, which are key concerns for practical implementation. Moreover, it is noted that each kind of existing decision-theoretic solution mainly addresses one aspect of the challenges, which implies that two or more kinds of decision-theoretic solutions should be incorporated to address more challenges simultaneously.