Abstract:

According to the non-stationary of frequency-hopping signals, a new blind source separation method based on time-frequency distribution is proposed. The method exploits the difference in the time-frequency signatures of the sources to solve the blind source separation problem based on the joint approximate diagonalization of a combined set of time-frequency distribution matrices of mixed signals. The theoretic analysis and simulation result shows that the new method is available for blind separation of frequency-hopping signals even if all the priori knowledge is unknown. Moreover, the proposed method has an increasing robustness with respect to noise.