Transient Interference Suppression And Target Detection Method In Over-the-horizon Radar

The problems of interference suppression and ship detection in over-the-horizonradar (OTHR) are always the research hot-spots, which face many technical difficulties.Among all interferences in OTHR, the transient interference has the great impact on theperformance of radar. The transient interference mainly includes the lightning in the air,the meteor echoes in the universe, artificial impulsive noise and so on. Although theadaptive digital beamforming (ADBF) can suppress these interferences to a certainextent, while there are still some remained interferences enter radar receiver. Theseremained interferences can lift up the noise bases and impact on the detectionperformance of the radar; even submerge the target into the noise background. So thetransient interference must be suppressed before ship detection. For OTHR, it is rathersuccessful in detecting aircrafts, owing to the full separation of sea clutter and thereturns of high-speed aircrafts in Doppler domain. However, it is difficult for the OTHRto detect ships on the sea surface. Firstly, the Doppler peaks of the ships are close to theBragg peaks of sea clutter, and lower than the Bragg peaks in amplitude (typically10-40dB). Secondly, due to the complication and the random fluctuation of thepropagation channel, the received signals in the OTHR suffer from the multipath andmultimode effects.OTHR operates in high frequency band and detects the targets by means of theionospheric refraction. Thus it works in a very complicated electromagneticenvironment. In the classical radar signal processing methods, the echoes are assumedto be stationary and the noise to be the white Gaussian. However, in OTHR, the echoesand noise are assumed to be stationary and white Gaussian, respectively. Due to theabnormal ionosphere environment, the echoes always deviate from the ideal scenarios.Consequently, it is necessary to study the new signal processing methods innonstationary, non-Gaussian environment. Empirical mode decomposition (EMD),which develops in recent years, is a very powerful method for nonlinear andnonstationary signal processing. Since OTHR echoes are typical nonstationary signals,the mainly work of the paper is about the application of the EMD in transientinterference suppression and ship targets detection. The summary of our work is asfollows:(1) Three new kinds of restraining interference in temporal domain are presented for theproblem of transient interference suppression of OTHR being solved. The first onecan effectively detect singularitiy of signal (transient interference) by using1-D discrete stationary wavelet transform, and the detail coefficient in wavelet influencecone are set to zero; finally,1-D inverse discrete stationary wavelet transform isused to the data reconstruction. In the second method, firstly, the complex empiricalmode decomposition (CEMD) is used to separate the radar echoes into severalintrinsic mode functions (IMFs), and then the first IMF determines the position ofthe transient interference, the correlation estimates the number of the IMFs whichneed to be eliminated; finally, the data is reconstructed. The third method ofsuppressing interference uses the wavelet packet transform (WPT) to decompose thesignal echo into a set of narrow band signals prior to the application of EMD. Then,each of narrow band signals is decomposed into a set of IMFs by CEMDdecomposition, and a screening process is conducted to remove unrelated IMFsfrom the result. At last, transient interferences are detected and excised in theresidual signal. The methods do neither require to perform clutter suppression, norto reconstruct data by using prediction algorithm of autoregressive (AR) model. Thus,these methods are of low complexity and high feasibility. Processing of the real datafrom OTHR system shows that these algorithms can efficiently suppress transientinterference.(2) A new sea clutter cancellation method based on CEMD is proposed for improvingthe capability of ship detection in OTHR. A number of stronger sea cluttercomponents are effectively extracted from the radar echoes by the CEMD and thenare cancelled. Thus, ship detection can be carried out. The experiments using realdata demonstrate the effectiveness of the proposed method for ship detection inOTHR.(3) A new model of sea clutter time series is founded under the multimode propagation.It considers a sea clutter time series as a multicomponent signal with severalnarrowband amplitude modulation frequency modulation (AMFM) components.The equal band complex empirical mode decomposition is used to obtain a stableDoppler spectrum of received time series based on the new model. It is firstdecomposed into several narrowband AMFM components by the CEMD and thenthe Fourier Doppler spectra of these components are superimposed to form theCEMD-based Doppler spectrum. For real data, the analysis and experiments showthat the CEMD-based Doppler spectra of sea clutter time series have much lessfluctuation in contiguous range cells and neighboring Doppler bins than the FourierDoppler spectra do. Embedding the CEMD-based Doppler spectrum into theconstant false alarm rate (CFAR) detection scheme in the range-Doppler domain yields a new hybrid CFAR detector for ship detection. The experiments show thatthe new detector attains better performance relative to the hybrid CFAR detectorusing traditional Fourier Doppler spectra.