A recursive estimation algorithm using selective updating for spectral analysis and adaptive signal processing

This correspondence presents a recursive estimation algorithm which, unlike conventional ones, updates the estimates only when a sufficient improvement can be obtained. With a bounded noise, assumption, the resulting sequence of estimates is a sequence of convex sets (ellipsoids) in the parameter space. For the cases studied, the algorithm used less than 20 percent of the data to update the estimates and still acquired very good accuracy for spectral estimation.