This paper investigates the problem of dynamic spectrum access for canonical wireless networks, in which the channel states are time-varying. In the most existing work, the commonly used optimization objective is to maximize the expectation of a certain metric (e.g., throughput or achievable rate). However, it is realized that expectation alone is not enough since some applications are sensitive to fluctuations. Effective capacity is a promising metric for time-varying service process since it characterizes the packet delay violating probability (regarded as an important statistical quality-of-service index), by taking into account not only the expectation but also other high-order statistic. Therefore, we formulate the interactions among the users in the time-varying environment as a non-cooperative game, in which the utility function is defined as the achieved effective capacity. We prove that it is an ordinal potential game which has at least one pure strategy Nash equilibrium. Based on an approximated utility function, we propose a multi-agent learning algorithm which is proved to achieve stable solutions with dynamic and incomplete information constraints. The convergence of the proposed learning algorithm is verified by simulation results. Also, it is shown that the proposed multi-agent learning algorithm achieves satisfactory performance.

In this study, the authors study the self‐organising user assignment problem for the multi‐user cooperation network. In the cellular network with user cooperation, idle users near the base station are seen as potential relay nodes, which can assist cell edge users to transmit data. Practically, the selfish nature of users is considered in the authors’ model. They design a score‐based system, where the strategies of relay assignment are affected by the historical behaviours of users. That means not only the transmission performance, but also the accumulative contributions of users are considered. The proposed score‐based system sufficiently encourages idle users to assist active users. Furthermore, the multi‐user assignment problem is formulated as a one‐to‐one matching game, in which idle users and active users rank one another individually based on their own preference. To address the issue, they propose a self‐organising mutually beneficial matching algorithm, which is proven to converge to a stable matching. Simulation results show that the proposed distributed algorithm yields well matching performance between source users and relay users, which is close to the optimal centralised results.

This paper investigates the problem of distributed channel selection for interference mitigation in cognitive radio networks (CRNs) with block-fading channels, using a game-theoretic solution. Specifically, the channel gains are blockfixed in a slot and change randomly in the next slot. Existing algorithms, which are originally designed for static channels, can not converge in the presence of time-varying channels. We formulate this problem as a non-cooperative game with random payoffs, in which the utility of each player (CR user) is defined as the expected weighted experienced interference. This game is proved to be a potential game with the network utility, the expected weighted aggregate interference, serving as the potential function. Then, we propose a stochastic learning automata based distributed channel selection algorithm, with which the CR users learn the desirable channel selections from their action-payoff history. It is analytically shown that the proposed learning algorithm converges to pure strategy Nash equilibrium (NE), which maximizes the network utility globally or locally, without information exchange. Moreover, simulation results show that it achieves higher normalized transmission rate.

This article investigates the problem of distributed channel selection in opportunistic spectrum access (OSA) networks with partially overlapping channels (POC) using a game-theoretic learning algorithm. Compared with traditional non-overlapping channels (NOC), POC can increase the full-range spectrum utilization, mitigate interference and improve the network throughput. However, most existing POC approaches are centralized, which are not suitable for distributed OSA networks. We formulate the POC selection problem as an interference mitigation game. We prove that the game has at least one pure strategy NE point and the best pure strategy NE point minimizes the aggregate interference in the network. We characterize the achievable performance of the game by presenting an upper bound for aggregate interference of all NE points. In addition, we propose a simultaneous uncoupled learning algorithm with heterogeneous exploration rates to achieve the pure strategy NE points of the game. Simulation results show that the heterogeneous exploration rates lead to faster convergence speed and the throughput improvement gain of the proposed POC approach over traditional NOC approach is significant. Also, the proposed uncoupled learning algorithm achieves satisfactory performance when compared with existing coupled and uncoupled algorithms.

This article investigates the problem of distributed channel selection for opportunistic spectrum access systems, where multiple cognitive radio (CR) users are spatially located and mutual interference only emerges between neighboring users. In addition, there is no information exchange among CR users. We first propose a MAC-layer interference minimization game, in which the utility of a player is defined as a function of the number of neighbors competing for the same channel. We prove that the game is a potential game with the optimal Nash equilibrium (NE) point minimizing the aggregate MAC-layer interference. Although this result is promising, it is challenging to achieve a NE point without information exchange, not to mention the optimal one. The reason is that traditional algorithms belong to coupled algorithms which need information of other users during the convergence towards NE solutions. We propose two uncoupled learning algorithms, with which the CR users intelligently learn the desirable actions from their individual action-utility history. Specifically, the first algorithm asymptotically minimizes the aggregate MAC-layer interference and needs a common control channel to assist learning scheduling, and the second one does not need a control channel and averagely achieves suboptimal solutions.