Analysis of multicomponent chirp signals using frequency-shear representation

This paper is concerned with the problem of multicomponent chirp signal analysis. The traditional analysis tool is time-frequency distribution which has been paid more attention in deterministic nonstationary signal processing. However, for chirp signals, it is difficult to find a best time-frequency representation that has high auto component concentration and cross-term suppression. Generally, any time-frequency representation based on rectangular resolution cell will not result in good performance for signals with time-varying frequency content. In this paper we propose a new representation for analyzing multicomponent chirp signals using resolution cell that shears in the time-frequency plane. In our approach, the kernel function with its shape adapted to the local structure of the chirp signal is used so that the signal components can be clearly mapped onto the frequency-shear plane. It is shown that the frequency-shear representation can reduce spurious values and concentrate on the auto components. Several signals including sinusoidal signals and chirp signals are analyzed and simulated using the proposed approach in comparison with the results by the Wigner distribution.