A Game-Theoretical Learning Approach for Spectrum Trading in Cognitive Satellite-Terrestrial Networks

Abstract

This letter investigates the bandwidth allocation problem for discontinuous frequency bands in cognitive satellite-terrestrial networks (CSTNs). Different from existing studies, we take the realistic assumption that the available idle spectrum bands are disconnected. Most of the existing works focused on the continuous bandwidth allocation problem and proposed corresponding algorithms optimizing continuous variables, which cannot directly solve the bandwidth allocation problem for discontinuous frequency bands. Besides, we also take the individual bandwidth requirement, the priority of terrestrial cognitive users, and the limitation on the total available bandwidth into consideration. The bandwidth allocation for discontinuous frequency bands is formulated as a bandwidth auction game, which is proven to have at least one pure strategy Nash Equalization (NE) solution. Furthermore, an improved best response learning algorithm based on exploration and exploitation (BREE) is proposed to achieve satisfying performance. Finally, simulations are conducted to verify the convergence and superiority of the proposed BREE algorithm.