Research Of Ultra-wideband Radar Signal Processing Based On Wavelet Theory

The ultra-wideband(UWB) radar is a sort of new radar system, which relative bandwidth has greater than or equal to 20% centre frequency. It has incomparable advantage to conventional radar such as high range resolution, low intercept probability, anti-interference, anti-stealth, anti-multipath and peculiar penetration. Most of research institutions are still at experimental study and theoretical research phase. In signal processing field, fourier transform is the main tool and in dominance; wavelet transform is its development and more and more used in signal processing field in recent years. According to the characteristics of UWB radar signal, this paper focused on two questions based on wavelet theory. One is wavelet two channels signal sampling system to reduce UWB signal sampling rate; another is wavelet de-nosing to improve SNR, namely improve signal detection probability.First, UWB radar signal is very short narrow pulses in time domain, and occupy very wide bandwidth in frequency domain. It makes signal acquisition difficult and often beyond the capability of the extended existing ADC devices. To reduce signal sampling rate this paper constructed a two channel sampling system by wavelet method. Signal is transformed to wavelet domain and its frequency band is divided into two equal parts by constructing orthogonal mirror decomposition filter bank and is reconstructed effectively by corresponding reconstruction filter bank.Secondly, UWB radar return signal is faint and time-varying, and detected difficultly. To improve signal detection probability this paper used wavelet method for de-nosing processing. Using good details analytical ability of wavelet in time-frequency domain, for threshold de-noising, it contrasted signal wavelet coefficients with noise coefficient to suppress noise, and quantitative analysis to the de-noising effect.Finally, by the way of theoretical derivation and software simulation the conclusions of this paper are as follows: (1) the orthogonal mirror filter bank constructed by wavelet can reduce sampling rate; (2) wavelet de-noising can effectively improve the SNR.