Eigenvalue-Based Spectrum Sensing Under Correlated Noise for Multi-dimensional Cognitive Radio Receiver

Spectrum sensing is a fundamental stage in cognitive radio networks. The eigenvalue-based spectrum sensing is an optimum blind sensing scheme for sensing of correlated signals. However, its performance severally degrades under correlated noise. The correlation in noise samples occurs due to oversampling and imperfection in filtering. Herein, to address such situations, the eigenvalue-based spectrum sensing is analysed using recent Random Matrix Theory (RMT) for multi-dimensional cognitive radio receiver. First, a detailed analytical analysis is carried out for the asymptotic distribution of maximum eigenvalue of sample covariance matrix of the received signal at secondary user. Then, a double-threshold based spectrum sensing technique is proposed, which improves the performance of single threshold-based sensing technique. The performance of the proposed double threshold-based spectrum sensing scheme under correlated noise is compared with existing eigenvalue-based single threshold spectrum scheme and conventional Maximum-minimum eigenvalue (MME) -based spectrum sensing technique in terms of probability of detection and required number of samples. The simulation results confirm the significantly improved performance of the proposed double threshold-based spectrum sensing technique under correlated noise over existing methods.