Analysis of eigenvalue based spectrum sensing techniques in cognitive radio networks

Abstract

Energy based spectrum sensing detection is optimal in terms of computational complexity butthey have certain limitations of their dependence upon noise. In contrast Eigenvalue basedalgorithms do not depend upon noise uncertainty. Eigenvalue based algorithms are computationally complex as compared to energy detection method. Its complexity comes from two steps, the decomposition of the covariance matrix and the computation of Eigenvalue. Thedecomposition of covariance matrix does not offer enough room for complexity analysis as it hasalready been studied to its maxima while the computation of Eigenvalues is still an open field forresearch.In this work we propose fast iterative algorithms to handle Eigenvalue problems forEigenvalue based spectrum sensing detections. The proposed algorithm reduces the complexityof the Eigenvalue based spectrum sensing techniques to . When the noise floor is highenough i.e. the signal is too weak its detection is challenging. The aim of the thesis is to detectweak signals in cognitive radios through varying thevaluesforthereceivedtimesample(smoothing factor) L with minimal complexity. Simulations based on real-time GSM signals andthe wireless microphone signals are presented to verify the proposed.We have reduced the overall complexity of the Eigenvalue based spectrum sensing techniqueswhich will be beneficial especially for cooperative spectrum field where we deal with multiplereceiver and transmitter signals. Most importantly the significance of work is in the detection ofweak signals in cognitive radios. As the signal becomes weak it will be smoothly detected usinglarger values for received time samples L. As L increase the complexity also increases where wecan use the proposed work in Eigenvalue based spectrum sensing methods to obtain sensingresults with reduced complexity.