Active Learning in Cooperative Communication Based Cognitive Radio

This paper considers the spectrum sharing (SS) between primary radio (PR) cooperative communication network and a cognitive radio (CR) link. An active learning method in cooperative communication based CR is proposed. Assuming PR cooperative node adapts its transmission power and rate based on the feedback from PR destination node, and three power adaptation policies of PR are investigated. CR transmitter adopts cooperation which CR transmitter cooperates to send the PR probing signal to PR destination, and observes PR transmission power and rate adaptations. We present the performances of channel power gain estimation with the active learning method, and give the lower and upper bounds of channel estimation. We also investigate the PR capacity loss with the proposed active learning method, and the estimation bounds of PR capacity loss are also given in this paper. Numerical results are provided to evaluate the performances of channel estimations with active learning under different power adaptation policies, and show that there is a tradeoff between the performance of channel gain estimation and capacity loss which should be considered in design.