A weighted ELRT-based robust spectrum sensing algorithm

Most of current spectrum sensing methods assume the probability density functions (PDFs) about some parameters are known beforehand. However, in most practical applications, the communication environments usually vary with time. The presumed PDF estimated previously might be different from the actual one. It will degrade the performance of most spectrum sensing methods dramatically. In this paper, we propose a robust approach to address this problem. The weighted Empirical Likelihood Ratio Test (ELRT) is an effective method in statistical mathematics which can improve the robustness against the effect of imprecise knowledge of PDF under small sample size. The weighted ELRT obtains the robustness by re-weighting the contributions of actual PDF to likelihood. We employ the weighted ELRT in spectrum sensing, and term this new method as weighted ELRT-based robust spectrum sensing. Simulation results corroborate the performance of the proposed method over those of conventional ones.