Sparsity and concentration measures for optimum quadratic time-frequency distributions of Doppler signals

Reduced interference distributions are well-known methods used for quadratic time-frequency signal analysis of nonstationary data including Doppler signals. These distributions use kernels that determine the quality of a time-frequency signal representation. Optimal reduced interference distribution kernels are determined based on the employed design criteria, including sparsity and concentration measures that are applied in the time-frequency domain. This paper demonstrates that sparsity, compared to common concentration measures, satisfies desirable performance attributes and, as such, is considered effective in revealing the signal local frequency structure and its time-frequency signature, including the signal instantaneous frequency.