Study On Some Problems Of Signal Processing In Buoy-based HF Surface Wave Radar

High frequency electromagnetic wave has the characteristic of small attenuation when propogating along the conductive sea surface. With this feather, the over-the-horizon detection can be achieved by high frequency surface wave radar (HFSWR). At present, shore-based mode is mainly adopted in HFSWR, but it is difficult to meet the detection requirement in open sea area since the limited detection range. Buoy-based HFSWR, using buoy as the platform of HF radar, can provide an alternative way to solve this problem. In addition to this, buoy-based HFSWR can be taken as nodes, together with the shore-based HF radars, to forming an HF radar network. This paper is based on the Twelfth-Five-Year Chinese High-Technology863Plan Project with the issue "Development of buoy-based high frequency surface wave radar system". The statements are discussed from the following aspects:signal modeling transient interference suppression, sea echo modeling and simulation, motion compensation:1. The system of buoy-based HFSWR and its signal processing framework are introduced. Except for the transform-receive system, there is a buoy platform subsystem to provide an ensurance of the whole system. The received signal is deduced within the twice-FFT signal processing framework. The motion of platform may cause disturbance of range between radar and scatters. And within the twice-FFT signal processing framework, the disturbance may result in phase contamination. If the intensity of disturbance exceeds a threshold, the Doppler echo spectrum may be shifted or broadened. Based on the basic mathematical model of array signal, the array signal model of buoy-based HFSWR is deduced, and influences of array signal processing such as digital beam forming and MUSIC (Multiple Signal Classification) algorithm by six degrees of freedom movement of buoy are anlysed.2. Transient interference is one of the main external interference in HFSWR, as well as in buoyed-based HFSWR. Its appearance will cause signal-to-noise ratio (SNR) dropping and affect the detection result. The overall framework of transient interference suppression is given:signal preprocessing-interference detection-surppression-signal recovery. In the part of interference detection and suppression, wavelet is utilized and a method based on S transform is proposed.which makes full use of time-frequency feather of the interference. Compared with the former method, the latter method avoids the step of sea echo suppression; in the part of signal recovery, two prediction methods based on AR model and BP neural network respectively are used to reconstruct sea echo. The validity and effectiveness of the methods are certified by measured data, and the SNR of Doppler spectrum can be improved for about5-10dB3. Scatter coefficient, namely scatter cross section is one of the main ways to describe sea echo characteristics. In buoy-base mode, due to motion of the platform, sea echo characteristics are quite different from that in shore-based mode. The back scatter coefficient in shore-based HF radar is introduced. Base on this, the back scatter coefficient in buoy-based radar is deduced. In this model, the motions are classified as two types:the first one is linear movement, which is caused by surface wind and current, expressed as uniform motion; the other is nonlinear motion, caused by swell and expressed as sine function. The Doppler spectrum is simulated using this model, and the effect under different conditions such as wind direction, wind speed, working frequency are analysed. Finally, the spectrum of wide beam radar is simulated.4. The influence of radar signals by buoy movements can be summarized to array error and the echo phase pollution. Array error involving amplitude error and phase error are caused by posture disturbance and can be expressed by analysis formula. So, array errors can be calibrated by posture parameter supplied by posture sensor. The calibration error is analysed with the posture parameter errors. Assuming the mesured parameter errors less than10%, and roll or pitch angles less than30°, the error factor of calibration is less than0.1. In the part of phase comtamination,4methods are introduced to solve the problem:motion-paremeter-based method, phase gradient method (PGA), maximum likelihood estimation method (MLE), and time-frequency analysis method based on S transform (ST). All of these methods are tested by measured and simulated data, and the results show the contaminated phases are improved after processing, the broadened Bragg peak sharpened again. The performance of the methods are analysed under different conditions such as SNR, types of contamination functions, parameters of contamination functions.