Bayesian Compressed Sensing Based Dynamic Joint Spectrum Sensing and Primary User Localization for Dynamic Spectrum Access

Abstract

In cognitive radio (CR) and dynamic spectrum access (DSA) network research, most of current work on spectrum sensing focuses on the detection of existence of the spectrum holes for secondary user (SU) to harness. However, in a more sophisticated CR, the SU needs to detect more than just the existence of primary users (PUs) and spectrum holes: e.g., the transmission power and location of the primary users. In our previous work, we combined the spectrum sensing and PU power/localization detection together, and developed a joint PU detection and power/localization detection algorithm via compressed sensing (CS). By employing compressed sensing, the measurement ratio for the spectrum sensors is significantly reduced. However, if the measurement ratio is too low, the compressed sensing algorithm will not provide accurate estimates. Since the sparsity in the frequency domain is dynamically changing, it is unfeasible to set a predetermined measurement ratio. In this paper, we extend our previous work to employ the Bayesian Compressed Sensing (BCS) to improve the reconstruction results and dynamically determine the measurement ratio. Specifically, the BCS algorithm provides an error bar along with the reconstruction of the target vector. This error bar can then be used to determine if the current measurement ratio is sufficient. When the spectrum environment changes, the error bar will change accordingly, giving us direction to increase or decrease the measurement ratio. Simulation results including the measurement ratio, the miss detection probability (MDP), false alarm probability (FAP) and reconstruction probability (RP) confirm the effectiveness and robustness of the proposed method.