Cooperative Spectrum Sensing Based on Complexity Measures for Cognitive Radio Communications

Over the past decade, distinguishing deterministic signal from noise in the radio frequency bands has become an active issue in the evolution of cognitive radio technology. To address this problem, numerous signal processing algorithms have been described based on entropy/complexity measurement. However, the methods that depend on finite empirical data lack connectivity between signal detection and appropriateness of the selected entropy technique. In this work, we investigate and characterize the chaotic nature of received signal based on Lyapunov exponents. After observing a seemingly chaotic nature, appropriate entropy measures are investigated to estimate the degree of chaosness in the signal. Based on comparative study, an optimal approximate entropy (OApEn) based detection technique is proposed. It estimates the optimal adaptive tolerance (r_t) using a heuristic search algorithm. The work is extended to cooperative sensing by introducing weighted gain combining based on fractal dimension (WGCFD) technique. The results reveal that the WGCFD with the OApEn algorithm can detect received signals up to - 22 dB with five nodes in cooperation. It outperforms the other detection methods frequently used for signal detection.