Linear prediction approach to oversampling parameter estimation for multiple complex sinusoids

The problem of oversampling parameter estimation for noisy sinusoidal signals is addressed. We first extend the weighted least squares (WLS) approach to the complex sinusoids. Then the oversampling weighted least squares (OSWLS) estimator is proposed based on data decimation. Estimation performance of the OSWLS method is analyzed via theoretical and simulation studies. Results are also compared to those of the WLS and decimative unitary ESPRIT methods as well as Cramér–Rao lower bound.

► We extend the WLS method for real tones to complex-valued sinusoidal measurements.
► We utilize the oversampling technique to improve estimation accuracy.
► The WLS algorithm with decimation alleviates the reduced frequency separation from oversampling.
► The estimation performance is analyzed and validated by computer simulations.