Study On Signal Processing Of Interference And Clutter For Sky-wave Over-the-horizon Radar

Radar, which plays an important role in the mission of warning, detection, tracking, identification and guidance in electronic warfare(EW), is called as clairvoyant on the battlefield. With the development in science and improvement in signal processing, modern radar is also developed in a direction of diversity and multi-system. In the different radar system, the sky-wave Over-the-Horizon Radar(OTHR) has aroused increasing academic interest for the unique over-the-horizon detection. The sky-wave OTHR, operating in high-frequency(HF) band(3-30 MHz), uses sky wave propagation by refracting from the ionosphere to detect the ships, aircraft and missiles and so on. The OTHR overcomes the line-of-sight limitation induced by the global curvature, thus the OTHR has a long detection range and is primarily useful to the early warning. At the same time, OTHR also has ability for detecting stealth targets and extreme low-altitude targets, confrontation with ant-radiation missile and integrated intelligence monitoring.Though the OTHR has many advantages, it still faces numerous challenges. The performance of the OTHR is seriously affected by external environment. Compared with the normal microwave radar, many complicated signal processing methods must be considered. Transient interference excision and ship targets detection have always been the focus of research in OTHR signal processing. Many scholars have used modern technology of signal processing on them and worked pretty well. The Low-rank Matrix Recovery(LRMR), which has been successfully applied on image signal processing and radar signal processing and so on, is widely researched and developed in recent years. In this paper, the LRMR is introduced in OTHR signal processing, and its application in transient interference excision, clutter suppression and ship targets detection have been researched and discussed. In the end, the paper has also studied the blind signal processing on the ship target detection. The summary of our research work includes the following sections:The first part mainly discusses about LRMR theory, which includes Matrix Completion(MC), Low-rank Representation(LRR) and Robust Principal Component Analysis(RPCA). A discussion on the RPCA is taken in the paper and some typical algorithms for RPCA have been studied in detail.The second part focuses on the transient interference excision in OTHR. By analyzing the types and characteristics of the transient interference, it is usually short in the slow time-domain, but the intensity of it is great. The received time-domain signal in a CPI for a given rang cell can be reconstructed as a Hankel matrix, in which the transient interference shows sparsely. According to the model of the received signal in the given range cell, the clutter signal and the target signal show the property of low rank in the Hankel matrix. Thus the theory of the LRMC can be applied on the transient interference excision, and a RPCA method which is used for complex data is proposed to suppress the transient interference. The proposed method is no need for transient interference location and data interpolation, which has the advantage for multi-transient interference and small energy transient interference suppression. The proposed method is also robustness on noise and then it has improved the SNR(Signal to Noise Rate) of the received signal, which is benefited for dim target detecting.The third part studies the clutter suppression in short coherent integration time(CIT) in OTHR. The received signal in OTHR always includes strong clutter signal, which is concentrated within only a short extent of Doppler bins around zero Doppler. However, the Doppler frequency of the slow moving target such as ship always has a low frequency and will be covered up by the strong clutter signal in the Doppler-domain. Thus the clutter should be suppressed before detecting the ship target. A method is applied on separation of clutter and target signal by increasing the CIT to have a good frequency resolution. However, it may introduce the problems such as improving the probability of ionospheric disturbances. Considering the characteristics of the substantial correlation between the adjoining range cells in the received signal, a novel method based on RPCA is proposed for clutter suppression in short CIT. A range-sweep map is reconstructed by adjoining range cells in the received signal and the clutter signal has the low-rank property in the matrix. The target signal always exists in one to two range cells, which makes it sparse in the matrix. Therefore, the clutter could be suppressed by separating the low rank signal and sparse signal with low rank matrix recovery.The fourth part explores the target detecting in OTHR under the incontinuous sampling mode which results in parts of signal missing. Less of the available methods have taken the sparse OTHR signal case into account. On the work of the previous one, a new method base on MC is proposed for target detecting. The target signal and the clutter signal has the property of low lank in the Hankel matrix reconstructed by the received signal, and then the lost data can be recovered by MC in time domain. The target can be detected by the reconstructed spectrum. Compared with the Compressed Sensing(CS) method, the proposed method shows better performance. And then the blind signal processing is introduced into the separation for clutter signal and target signal, which could be applied on target detection under the background of clutter signal. However, the number of the target and clutter is unknown in the received signal, and they also may be non-disjoint in both time and frequency domain. The traditional blind signal processing methods do not work well on it. In this paper, a canonical decomposition approach for underdetermined blind separation of non-disjoint sources is proposed, with which the target signal can be separated from the received signal in OTHR.