Blind Multiband Joint Detection in Cognitive Radio Networks Based on Model Selection

Multiband joint detection (MJD), which simultaneously detects the primary signals over multiple narrow spectrum bands and aggregate the vacant channels for secondary transmission, is taken as an essential implementation of wideband spectrum sensing for efficient cognitive radio networks. In this paper, we propose a simple but effective sensing scheme to implement blind MJD. Our scheme can be explained as an improved implementation of energy detection based on model selection. By exploiting eigenvalue analysis technique developed from random matrix theory (RMT), our method first estimates the number of occupied channels K̂, and then judges the K̂ channels with maximum power to be occupied by the primary users. Unlike conventional energy detection or information theoretic criterion based sensing methods, our approach doesn´t require the knowledge of noise power or primary signal distribution, and thus is more suitable for blind spectrum sensing. Both theoretical analysis and simulation results validate the efficiency of the the proposed scheme.