Application Of Adaptive Subspace Detection In High Frequency Surface Wave Radar Ionosphere Clutter

Taking advantage of high frequency vertical polarization electromagnetic wave which can propagate circularly on the surface of sea, high frequency (HF) gorund wave over-the-horizon (OTH) radar can detect ships and aircrafts at low altitude from a strong combined background of sea echo, external noise, and interference. However, the wave reflecting from ionosphere will always enter into receiving systems, disturbing receiving signal strongly, and enhancing the base, reducing signal-to-noise ratio. Thus, it makes a lot of sense to suppress the interference from ionosphere. Adaptive subspace detection algorithm is proposed in the background of Gaussian distribution. It suppresses interference space which is not matched to the signal space while the information in the signal space is retained. This paper tries to use adaptive subspace detection algorithms to reach the aim of suppressing the ionosphere clutter.First, we model the signal of sea clutter, target and interference received by radar receiving systems, and model ionosphere clutter by compound-Gaussian distribution. We then introduces generalized likelihood ratio test (GLRT) and maximum likelihood estimate, and make a proof for constant false alarm rate (CFAR) of GLRT.We introduce a detector matched space detector (MSD) based on GLRT. According to four kinds of different unknown information background, we obtain four different kinds MSDs. We give the computing process of detection statistics of each kind of MSD, and work out the performance curves of each detector and compare the performance change of detectors under different arrays, different number of interference.Then, we take a throughout analysis and research on the adaptive version of MSD—adaptive subspace detector (ASD), substituting the unknown noise covariance matrix by sample covariance matrix, and giving detection statistics of each adaptive detector. We get the decomposition of each statistic by five different statistics and work out the probability dense function of each detector. We compare the performance change of each detector with different number of sample and thus obtain the conditioned distribution function of each detector. Working out the graph of probability of detection with SNR changing under different number of sample, we compare the performance of three detector—Kelly, AMF and ACE in different situation.At last, we use adaptive subspace detector (AMF and CFAR ASD) to deal with radar simulation data and get the curves of Pd—Pfa under orthogonal and non-orthogonal of interference space and signal space. We also extract the adaptive weigh vector that can suppress the clutter from AMF and CFAR ASD after analyzing AMF and CFAR ASD, and prove the efficiency of this two methods. Then we propose a new suppress method—second time adaptive clutter suppression based on this two methods. After using it to deal with the simulation data, we test the effect of this method and extract target from ionosphere clutter efficiently.