Time Delay Estimation Based On Fourth Order Cumulants

Time delay estimation is an active research field in signal processing. It has various and important applications in the fields of radar, sonar, communications and so on. Great effort has been made about time delay estimation. With the advantages of blindness to Gaussian noise, high-order cumulants has been widely applied in time delay estimation in recent years. Research on time delay estimation under correlated noise based on high-order cumulants is put forward based on this idea.Previous work in least mean square time delay estimation based on fourth order cumulants (for short FOC—LMSTDE) owns the peculiarities of less computation and simple principium. The method only fit integer time delay estimation. To non-integer time, the method must come together with other methods. And we can observe that the method applies to the condition that the signal and the noise power is invariable. When signal and noise power is variable, the result is degradation in system performance, and we also can see that the ability of tracking has been weakened.To resolve the problem of FOC—LMSTDE algorithm that must come together with other methods to estimate non-integer time, the method of explicit time delay estimation based on fourth order cumulants (for short FOC—ETDE) has been proposed. It can directly estimate integer time delay and non-integer time delay. Because it can directly iterate time delay. When signal and noise power is variable, the result is degradation in system performance.To resolve the problem of FOC—LMSTDE algorithm that the ability of tracking has been weakened in variable SNR. The algorithm of explicit time delay and gain estimator based on fourth order cumulants (for short FOC—ETDGE) has been brought up. The single adaptive filter in the conventional model is separated into two adaptive subunits connected in cascade, one is to track the delay, the other to track the SNR. Comparison and analysis with the FOC—ETDE and the FOC—LMSTDE, when signal and noise power is variable or low,simulation results show that the method proposed has better performance because of the time delay and SNR separation.