Cepstrum based non-parametric approach in the context of spectrum hole search in cognitive radio network

Cognitive radio finds the solution of bandwidth constraints on the communication system and dynamically captures the unoccupied spectrum by a technique of spectrum sensing. Periodogram based spectrum sensing may be a valid technique where the power spectral density of the received signal can be estimated from the observed samples. However, the estimate may suffer from inconsistency problem if the sample size is large. To overcome this difficulty, the concept of cepstrum estimator has been coined in different literature where the estimator can be calculated from the logarithm of the estimated power spectral density. Also the inconsistency problem can be avoided by this method. In this article, the utility of cepstrum has been particularized in order to find vacant frequency bands. The performance of the detection technique has been justified with false alarm and detection probabilities. Simulation results are provided in support of our proposition.