A new wavelet-based algorithm for filtering low SRN signals

In this work we present a new algorithm suitable for detecting and filtering radar or sonar pulses in presence of strong noise, jamming signals and Doppler frequency shift in the received signal. The novel algorithm is based on the multiresolution wavelet analysis, a powerful tool right to analyze non stationary signals. We exploit the Wavelet-based time-frequency domain analysis to detect and reconstruct radar or sonar pulses, obtaining very good matching between transmitted and received signals up to a signal-to-noise level of -11 dB.

INTRODUCTION

Due to the severe constraints whether on the communication environment, or spectral images of biological tissue as well as vector images of blood flow, digital signal processing of sonic and ultrasonic waveforms often requires the use of advanced denoising filtering, in order to reduce the presence of noise and possibly false jamming signals. For example, in real time radar target detection echoes signals are elaborated by the system to extract useful informations [1] as the target distance and relative radial speed. Moreover, in some applications such as biomedical tissue microstructure reconstruction, material properties investigations or microwave automotive radar systems, the received echoes are so strongly corrupted by interfering signals that the system is not able to extract the information [2] [3] [4]‘ so it is necessary a reliable method to recover the objects' echoes.