Adaptive Games for Agile Spectrum Access Based on Extended Kalman Filtering

Abstract

For dynamic spectrum allocation (DSA), distributed game has emerged as an attractive approach that enhances radio spectrum utilization efficiency at scalable complexity in network size. Cognitive radios act as game players to judiciously decide their transmission power spectrum density (TPSD) based on channel state information (CSI), which is generally estimated independently from the DSA games and may cause large computational and communication overheads. Particularly in the presence of doubly selective fading channels, a conventional DSA game needs to re-train the channel estimator and re-calculate the TPSD decisions for every transmission burst. To enable DSA intelligence at affordable costs, this paper proposes novel adaptive DSA algorithms based on channel tracking. Under the framework of extended Kalman filter (EKF), both the unknown CSI and the TPSD are modeled into a state vector that is to be tracked dynamically. This approach leads to EKF-based adaptive games that jointly track the CSI and update TPSD decisions, resulting in fast convergence and reduced communication overhead. Simulations are performed to testify the effectiveness of the proposed DSA algorithms, in terms of the achieved system spectrum efficiency, communication overhead, as well as resilience to user mobility