Non-convex distributed power allocation games in cognitive radio networks

Resumen

In this thesis, we explore interweave communication systems in cognitive radio networks where the overall objective is to maximize the sum-rate of each cognitive radio user by optimizing jointly both the detection operation based on sensing and the power allocation across channels, taking into account the influence of the sensing accuracy and the interference limitation to the primary users. The optimization problem is addressed in single and multiuser cognitive radio networks for both single-input single-output and multi-input multi-output channels. Firstly, we study the resource allocation optimization problem for single-input single-output single user cognitive radio networks, wherein the cognitive radio aims at maximizing its own sum-rate by jointly optimizing the sensing information and power allocation over all the channels. In this framework, we consider an opportunistic spectrum access model under interweave systems, where a cognitive radio user detects active primary user transmissions over all the channels, and decides to transmit if the sensing results indicate that the primary user is inactive at this channel. However, due to the sensing errors, the cognitive users might access the channel when it is still occupied by active primary users, which causes harmful interference to both cognitive radio users and primary users. This motivates the introduction of a novel interference constraint, denoted as rate-loss gap constraint, which is proposed to design the power allocation, ensuring that the performance degradation of the primary user is bounded. The resulting problem is non-convex, thus, an exhaustive optimization algorithm and an alternating direction optimization algorithm are proposed to solve the problem efficiently. Secondly, the resource allocation problem for a single-input single-output multiuser cognitive radio network under a sensing-based spectrum sharing scheme is analyzed as a strategic non-cooperative game, where each cognitive radio user is selfish and strives to use the available spectrum in order to maximize its own sum-rate by considering the effect of imperfect sensing information. The resulting game-theoretical formulations belong to the class of non-convex games. A distributed cooperative sensing scheme based on a consensus algorithm is considered in the proposed game, where all the cognitive radio users can share their sensing information locally. We start with the alternating direction optimization algorithm, and prove that the local Nash equilibrium is achieved by the alternating direction optimization algorithm. In the next step, we use a new relaxed equilibrium concept, namely, quasi-Nash equilibrium for the non-convex game. The analysis of the sufficient conditions for the existence of the quasi-Nash equilibrium for the proposed game is provided. Furthermore, an iterative primal-dual interior point algorithm that converges to a quasi-Nash equilibrium of the proposed game is also proposed. From the simulation results, the proposed algorithm is shown to yield a considerable performance improvement in terms of the sum-rate of each cognitive radio user, with respect to previous state-of-the-art algorithms. Finally, we investigate a multiple-input multiple-output multiuser cognitive radio network under the opportunistic spectrum access scheme. We focus on the throughput of each cognitive radio user under correct sensing information, and exclude the throughput due to the erroneous decision of the cognitive radio users to transmit over occupied channels. The optimization problem is analyzed as a strategic non-cooperative game, where the transmit covariance matrix, sensing time, and detection threshold are considered as multidimensional variables to be optimized jointly. We also use the new relaxed equilibrium concept quasi-Nash equilibrium and prove that the proposed game can achieve a quasi-Nash equilibrium under certain conditions, by making use of the variational inequality method. In particular, we prove theoretically the sufficient condition of the existence and the uniqueness of the quasi-Nash equilibrium for this game. Furthermore, a possible extension of this work considering equal sensing time is also discussed. Simulation results show that the iterative primal-dual interior point algorithm is an efficient solution that converges to the quasi-Nash equilibrium of the proposed game.