ABSTRACT
Cognitive radio (CR) system plays a major role in spectrum sensing. In cognitive radio, the spectrum slot is assigned to user intellectually according to a prescribed algorithm. Different researchers have proposed different spectrum sensing methods to plot the receiver operating characteristics (ROCs), probability of detection (P D), and probability of false alarm (P fa). Apparently, there is less focus on which samples of P fa are majorly affected. A method of Hybrid Filter Detection with Inverse covariance matrix (HFDI) is proposed to identify the P fa affected samples and to minimize the P fa with maximize of P D. The proposed HFDI method uses generalized log likelihood ratio test (GLRT) to measure the P fa and P D. The measured P fa and P D are compared with the NP observer approach, HMF, and Divya Joshi-proposed methods. Additionally, a parameter probability of missed detection (P md) is measured between the proposed HFDI and NP observer approach. The proposed HFDI method outperforms Divya Joshi-proposed method by 13% in terms of P D, and the P fa is also improved by 7.5%. ROC is a parameter used to analyze the performance between P D and P fa and is compared with the existing compressed sensing method.