Autocepstrum Approach for Spectrum Sensing in Cognitive Radio

This paper provides a novel blind spectrum sensing technique based on cepstral analysis in interweave cognitive radio (CR) system. The main scope of this work is to mitigate the problem of weak signal detection so as to allow for interference-free spectrum sharing. The misdetection problem of a possible legitimate user occupies a desired frequency band leads to erroneous sensing results. Based on the periodicity revealing property of cepstral analysis approaches, we formulate a spectrum sensing technique based on the autocepstrum concept. We employ the proposed approach to detect a spread spectrum (SS) primary user (PU) signal. The blind theme of the proposed approach implies that no knowledge of the spreading code employed in a SS signal is provided at the CR receiver. The distribution of the detection test statistic is derived under the null hypothesis based on Neyman-Pearson lemma (NPL). The corresponding detection threshold is analytically computed. The performance of the proposed spectrum sensing algorithm is compared with conventional energy detection (CED) in terms of detection probability. As a result, the proposed detector outperforms CED, and indicates lower misdetection probability in low signal-to-noise (SNR) ratio environment.