Comparative Tracking Performance of SSRLS and SSLMS Algorithms for Chirped Signal Recovery

This paper compares the tracking performance of state space recursive least squares (SSRLS) and state space least mean square (SSLMS) algorithms for a chirped signal buried in additive white Gaussian noise. The signal is a sinusoid whose frequency is drifting at a constant rate. After incorporating second order linear time varying state space model of the chirped sinusoid into their formulation, both SSRLS and SSLMS exhibit superior tracking performance over standard RLS & LMS and their known variants. The performance comparison is based on the evaluation of time average auto-correlation function (ACF) of prediction errors of SSRLS and SSLMS when responding to the chirped signal for different values of forgetting factor (SSRLS) & step-size parameter (SSLMS). Relative whiteness of prediction errors of SSRLS and SSLMS gives a measure for comparing their tracking performance. Tracking results for standard RLS and LMS are also reported