Analyzing non-stationary signals: A joint time-frequency and spatial-frequency approach

Abstract

We present a robust non-stationary signal analysis tool that is applied to accurate parameter estimation of multi-sensor non-stationary signals. This tool analyzes multi-sensor signals in a joint Time–frequency and Spatial-Frequency (JTFSF) domain using quadratic and linear representations. It is illustrated that multi-source signals with overlapping Time-Frequency signatures become non-overlapping in JTFSF domains thus allowing more refined signal analysis than conventional Time-Frequency representations. The JTFSF domain representation is applied to solve the parameters (Instantaneous Frequency) estimation problem of closely placed components in the Time-Frequency domain by employing joint Time–Frequency and Spatial filtering. It is illustrated that the proposed tool achieves superior performance than the state of art methods in terms of accuracy.