An improved algorithm of PAST and Root-MUSIC for signal frequency tracking

For multi-signal separation and frequency tracking problems, this paper conducted a study and simulation of PAST and Root-MUSIC algorithm, and then pointed out that: due to the signal noise and subspace approach method of PAST algorithm, the solutions may contain many disturbance items. The original Root-MUSIC algorithm can not effectively eliminate the disturbance. This paper presents an improved method based on the tolerance and sorting. It divides all possible solutions into several neighborhoods, then puts every neighborhood for shrinkage and contraction to the local optimal solution, finally obtains the original signal frequencies by weighted and sorted processing. The results of various sources demonstrate the accuracy and the effectiveness of the improvement.