Frequency estimation based on linear prediction

Abstract

The aim of the project is to develop a computationally simple frequency estimation method based on a modified linear prediction technique. Its performance will be contrasted with other conventional approaches for different applications such as speech coding and power line monitoring. The proposed estimator with be tested with the existing estimators with different length of samples, signal to noise ratio, frequencies, computational complexity and different types of noise through computer simulation using MATLAB. Moreover, the proposed estimator will be extended to the signal model with damping ratio. The new estimator will be tested with different signal to noise ratio, length of samples, frequencies and damping ratio. Furthermore, signal consists of two sinusoids will be investigated and the performance of estimations by different estimators will be compared.