Cognitive radio networks: Learning, games and optimization

With the increasing demand for wireless communication, efficient spectrum management and access are necessary and critical. What we have witnessed is an underlying trend of paradigm shift from fixed centralized communication to decentralized and dynamic communication with highly distributed resources, computation and processing. With the Federal Communications Commission's (FCC) spectrum policy reform, new ways to dynamically access spectrum are becoming possible. This thesis addresses emerging problems in enabling cognitive radio networks and dynamic spectrum access through methods involving stochastic learning, game theory and optimization.; In the first part of this thesis, we consider some of the enabling technologies for dynamic spectrum access. Intra-network distributed power allocation with discrete power levels is studied. Each user's evaluation of the quality of service (QoS) is computed using a utility function which corresponds to the number of correctly received information bits per unit energy. Two stochastic learning algorithms are proposed to find the Nash equilibrium solution of this power allocation problem in a distributed fashion with limited information exchange.; We also study inter-network spectrum sharing. The problem of how to achieve fair and efficient sharing of radio resources between dissimilar radio systems that can not communicate with each other in unlicensed frequency bands is investigated. A random spectrum access algorithm is proposed to achieve optimal spectrum utilization and maintain access fairness. To guarantee that the secondary spectrum access users will never interfere with the primary users and at the same time be provisioned with a target signal to interference ratio (SIR), we present a novel QoS constrained secondary spectrum sharing scheme based on the concept of interference temperature (the total allowable interference in a spectral band). A distributed joint coordination and power control algorithm is developed to implement the secondary spectrum access. In this way, secondary users can coexist with primary users as long as the interference temperature constraint is not violated.; In the second part of this thesis, we consider the economics of emerging cognitive radio networks. We explore the price dynamics in a competitive market consisting of spectrum agile network service providers and users. Here, multiple self interested spectrum providers operating with different technologies and costs compete for potential customers. We use game theory to find the best response solutions for different providers with both quality sensitive and price sensitive user populations. A stochastic learning based strategy is used by the providers to set the price when the market information is limited.