Comparison of various periodograms for single tone detection and frequency estimation

Periodograms are a simple and efficient way to detect and estimate parameters of sinusoidal signals. In this paper, we attempt to determine the optimum periodogram implementation for a single real tone detection in the presence of white Gaussian noise, among its variants such as the Bartlett's procedure and Welch method. The possibility of delaying threshold effect using different realizations for frequency estimation is also investigated. Simulation results show that the standard periodogram generally gives the best detection performance and it provides the minimum mean square frequency estimation error for all signal-to-noise ratio conditions.