Harmonic Retrieval from Coarsely Quantized Measurements

We consider the problem of harmonic retrieval from coarsely quantized measurements. We formulate the sinusoidal parameter estimation problem into a sparse parameter estimation problem by adding an l q -norm (0 < q ≤ 1) based penalty term to the negative log-likelihood function. A user parameter-free algorithm is devised to solve the problem efficiently by using the cyclic optimization and majorization-minimization (MM) techniques. Numerical examples are provided to demonstrate the estimation performance of the proposed algorithm.