A Gradient Ascent Approach for Multiple Frequency Estimation

This work investigates a new approach for frequency estimation of multiple complex sinusoids in the presence of noise. The algorithm is based on the optimization of the least squares (LS) cost function using a gradient ascent algorithm. The paper studies the performance of the proposed method and compares it to other estimation techniques such as root-multiple signal classification (root-MUSIC) and the discrete-time Fourier transform (DTFT). Simulation results show the performance gains provided by the proposed algorithm in different scenarios.