Detection Based On The Rényi Entropy Of Time-Frequency Distribution

Detection is one of the most important aspects of signal processing. The detection methods based on time-frequency distribution have a wide research as the development of time-frequency distribution theory. In most of these methods, the accumulation of the signal energy from noisy observation is a key to develop an effectual detector. However, the detect methods based on the energy accumulation are invalid when the SNR is low. In this paper, exploring the information content of the time-frequency distribution of interesting signal, we develop a new detect method based on the Rényi entropy of the time-frequency distribution. Considering the influence of different distribution function of time-frequency distribution to the entropy, a new modified detect algorithm is present based on the Rényi entropy of Adaptive Optimal Kernel (AOK) time-frequency distribution to substitute the detect algorithm based on the Rényi entropy of Wigner distribution. Simulations show that the detect method based on Rényi entropy of time-frequency distribution has excellent performance when the SNR is low, and the modified detect method based Rényi entropy of Adaptive Optimal Kernel time-frequency distribution has much superiority over the detect method based Rényi entropy of Wigner distribution especially when the interesting signals are multi-component signals.