Cooperative sensing based on permutation entropy with adaptive thresholding technique for cognitive radio networks

Spectrum sensing in the low signal-to-noise ratio (SNR) environment is vital task for the evolution of cognitive radio technology. The numerous signal processing algorithms have since been proposed to improve the spectrum sensing performance. In the recent past, entropy based sensing methods are shown to be robust in a low SNR environment with small data sets. However, these methods only focus on information content and ignore temporal order of the signal. Hence, selection of appropriate entropy technique that considers both information content and temporal order is important. In addition, many works consider that the distribution of noise follows Gaussian under assumption that the sample size is infinity. The detection threshold designed using this assumption yield unreliable decisions. On the contrary, the captured data is limited in real-time and it should be minimum to reduce the computational complexity. To address these two issues, empirical permutation entropy with adaptive thresholding detection technique is proposed. Then, the work is extended to weighted gain cooperative sensing that uses Higuchi fractal dimension method to generate weight for each node. Simulation results reveal that the proposed method is robust, less sensitive to sample size, and improves the single node as well as multinode sensing performance.