Parameter estimation of LFM signals

Abstract

Linear frequency modulated (LFM) signal is widely used in many areas including sonar, radar and communication system. Many methods have been proposed in revealing the joint time-frequency characteristics of the LFM, such as, short-time Fourier Transform (STFT), Wigner-Ville distribution (WVD) and Radon-Wigner transform (RWT). However, these methods have some drawbacks of low time and frequency resolution, interference of cross terms and limited accuracy. A new method was introduced called Lv’s distribution (LVD) which is powerful tool for LFM signal estimation and detection. LVD is able to provide higher concentration of auto terms compared with WVD and has higher time and frequency resolution than STFT. However, some of the properties are not well studied and thorough comparisons with STFT are needed. This thesis studies the principles of LVD including the derivation and analysis of some of the important properties. And a thorough analysis and comparison of the LVD with STFT are made from both the theoretical and simulation perspectives, where the relationship and differences between the two methods are revealed. Numerical simulations are conducted to verify the theoretical derivations and analysis of LVD. The obtained results by using a multi-component LFM signal show that the parameter estimations of LVD are accurate even under low signal-to-noise ratio (SNR) environment and resolution given by LVD is much higher than STFT.