Research On Stap-Based Ionospheric Clutter Suppressing Method In High Frequency Surface Wave Radar

High-frequency surface wave radar (HFSWR) is designed to measure the sea state and to detect and track maritime targets beyond the horizon. This is achieved by the propagation of the vertically polarized electromagnetic surface wave and relatively low attenuation of these waves over the ocean surface. It offers numerous advantages for low-cost, real-time and weatherproof. However, target detection and tracking is primarily limited by ionospheric interference. The thesis will use space-time adaptive processing(STAP) to suppress ionospheric clutter, which start from studying of full space-time adaptive processing, and then lead to reduced dimension STAP processor-JDL algorithm and its improved algorithm, whose performance is shown by the simulation.Firstly, the theory of full STAP processor is introduced, depending on the interference covariance matrix eigenvalue decomposition, the impact of gain and phase uncertainties and the residual error of self calibration on STAP algorithm is proved. Addressing the shortcomings of SMI solution, which directly uses the interference covariance matrix's maximum likelihood estimation, an improved algorithm with low complexity-CMT's has been established, and simulations validate the performance of the algorithm. At last, the difficulty of using full STAP processor to the HFSWR is studied, and the feasibility of reduced dimension STAP algorithms for HFSWR is given.Finally, we attempt to implement reduced dimension STAP algorithms on HFSWR system. The concept of local clutter DOFs (degrees of freedom) is given and characteristics of ionospheric clutter's local DOFs are analyzed. Then, the JDL algorithm is introduced, and clutter suppressing experiments using the ideal point target and real target have proved JDL algorithm's great capability of suppressing ionospheric clutter. At the same time, applying CMT's into the original JDL algorithm, the paper proposes an improved algorithm which is so called JDL-CMT algorithm. It has better performance even using much less training data. The results also illustrate that careful selection of training samples and the local processing region (LPR) are very important to obtain good performance, as well as JDL algorithm's limitation on clutter non-homogeneity.