Detection Performance of MEM Processor in Radar Applications

Abstract

Fast Fourier Transform (FET) technique has been in use for processing the radar signals in doppler domain. The introduction of Miximam Entropy Method (MEM) by Burg [2] has revolutionized the field of non-linear spectral estimation. Since MEM processor can yield high resolution spectral estimates even with short data lengths, it is becoming a goad candidate for doppler domain processing of radar signals. In this paper, we obtain probability density functions for MEM spectra and show that they closely follow log-normal density function. The usual detection performance criteria i.e., probability of detection (P_d) and probability of false alarm (P^fa) are then obtained using log-normal density function. The detection performance of MEM processor is compared with that of FFT processor. It is observed that the MEM processor can yield better results compared to FFT processor when the doppler frequency of the radar return from a target does not coincide with the center frequencies of the FFT filter bank.

Indexing terms:

• Radar detection
• Signal processing
• MEM Processor