An open-loop adaptive filter for rapid frequency estimation

This paper describes an open-loop adaptive filter and algorithm which can be used for frequency estimation of non-stationary signals. The filter is free from stability problems and can deliver a rapid response for signals which contain no or low amounts of noise. The adaptive speed of the algorithm is not affected by changes in the amplitude of the input signal. This algorithm is well-suited to a fast, yet accurate, method of obtaining the frequency from the model weights which relies on the weights being linear phase. The performance of the frequency extraction method is illustrated.