Study On Nonhomogeneous STAP Methods

Conventional statistical STAP methods are usually studied for sidelooking uniform linear array (ULA). Under this condition, STAP algorithms can get the best performance because there are sufficient training samples with interference that are independently and identically distributed (IID) with interference in test sample to estimate covariance matrix. But for the the different array structure or the gemotry of the array, the distribution of the clutter will change with range and training samples and test sample are no longer homogerous, so statistical STAP methods degrade dramatically. Two typical nonhomogeneous STAP problems for airborne phased array radar:Uniform circular array (UCA) STAP and non-sidelooking uniform linear array (ULA) STAP are studied in this dissertation and the main work of this paper can be summarized as following:1. Clutter model of UCA is set up and the clutter characteristics of UCA are analyzed from the perspective of the distribution of clutter spectrum, distribution of eigenvalue and signal to clutter and noise loss. The conclusion that the non-linear relations between the space angle and the array elements result in the clutter spectrum's nonhomogeneity is obtained.2. Localized processing STAP algorithms for UCA are studied. Several typical localized processing C-STAP methods, such as 3DT,F$A, PC and CSM are analyzed and a new localized algorithm, namely RD-GSM method which combimes structured reduced dimension method with adaptive reduced rank method, is proposed. Simulation results show the proposed method outperforms significantly 3DT and F$A while the computation load is about the same as these two methods.3. The DDD method for UCA is studied and a modified DDD method is proposed. Comparied with conventional DDD method, only amplitude variation with fixed phase for weights connected with the space-time data of the modified method reduce the computational load about 50%, by implementing multiple space-time constraints, system gain is maintained when the signal of interest arrivals slightly offset in angle, Doppler, or both.4. Range dependence compensation algorithms for UCA are studies.In localized compensaition technologies, SAC and MADC methods are proposed to bring the clutter spectrum in mainlobe area together. In order to reduce the residual dispersion in the sidelobe, MADC-DBU and MADC-EDBU algorithms are presented to realign the angle-doppler trajectory around the mainlobe after the spectrum centers in different cells have been co-located. Simulation results show these two stages hybrid methods are able to further reduce the clutter dispersion with the respect to localized compensation methods, but the computation load increases greatly. 5. Linear Prediction STAP methods for UCA are studied. Two typical linear forcasting methods (C-STAR and C-PICM) are analyzed and simplified parameters determination methods are adopted. Simulation results show C-STAR outperforms localized compensation techniques at the same computation load while C-PICM is sensitive to the accuracy of the inverse matrix sequences in the surrounding range getes.6. Range dependence compensation techniques for non-sidelooking ULA are studied and Multiple Space Angle Compensation (MSAC) algorithm is proposed to accomplish space angle alignment at multiple Doppler frequencies. Compared with DW and ADC methods, MSAC algorithm can further reduce the geometry-induced nonstationarity while the computation load increases only a little.