Robust multiple hypothesis testing approach for dynamic spectrum sharing

This paper tackles the four-level hypothesis problem for spectrum sharing between primary users (PUs) and secondary users (SUs). The considered hypotheses are as follows; the studied frequency band is vacant (hypothesis H₀), the studied frequency band is occupied only by a PU (hypothesis H₁), the studied frequency band is occupied only by another SU (hypothesis H₂) and the studied frequency band is occupied by PU and another SU simultaneously (hypothesis H₃). Most of the related works have considered the binary hypothesis testing problem (H₀ and H₁). To enable four-level hypothesis testing, in our previous work we used cyclostationary based spectrum sensing with intentional frequency shift in the SU signal. We found the minimum frequency shift that leads to a required level of detection performance. However, random carrier frequency offset due to a mismatch between oscillators has not been considered even though it may deteriorate the detection performance significantly in practical systems. Therefore, we propose a robust cyclostationary spectrum sensing technique. In addition, in the multiple hypothesis problem, setting design parameters to satisfy constraints is not easy. Therefore, we propose a method to set proper design parameters which can satisfy target detection and classification error probabilities. Numerical result verifies a validity of the proposed parameter setting method.