GOODNESS-OF-FIT BASED SPECTRUM SENSING USING EXPONENTIAL DISTRIBUTION IN LAPLACE NOISE

Spectrum sensing for cognitive radio in non-Gaussian noise is challenging. Most existing literatures on spectrum sensing only consider the Gaussian background noise, which is not always valid in practice. Motivated by thesis, a spectrum sensing scheme based on the order statistics goodness-of-fit (GoF) test of the absolute value of the received samples against an exponential distribution in a Laplace noise environment, is proposed. The detection performance of the proposed algorithm is evaluated using Monte Carlo simulations. We compute the receiver operating characteristic (ROC) of the proposed method and show its superior performance to detect a random signal under Laplace noise, compared to existing GoF tests using Anderson-Darling, likelihood ratio and energy detection methods under the same conditions. Especially, in a Laplace noise environment achieves higher performance with short sensing interval.