Tarset Detection And Localization Via MIMO Synthetic Aperture Sonar Processing

For the detection and localization of acoustic targets, one of the most effective ap-proaches is through Synthetic Aperture Sonar (SAS) imaging. The conventional monos-tatic SAS imaging is based on the point target assumption, where the fluctuation of the Target Scattering Patten (TSP) within the synthetic aperture is not considered, and only backscatter echo from the target is used. As to a distributed target, the target fluctuation is a nuisance parameter to the performance of an active sonar, which can not be handled by coherent accumulation through conventional SAS processing. On the other hand, the synthetic aperture size is constrained by the limited width of the main lobe of TSP.By placing the transmitter and the receiver at different positions, a bistatic sonar makes observations of the target from different points of view, acquiring angular dependence of TSP. This can be a big advantage over a monostatic sonar, provided that the TSP of acoustic targets are well modeled. Besides, both of bistatic sonar and Multiple Input Multiple Output (MIMO) sonar embody the concept of "Multi-", admitting the existence of spatial fading of acoustic targets and trying to make use of it.This thesis is titled as "Target detection and localization of targets via MIMO-SAS processing", and concerns some key problems when the SAS technology is faced with dis-tributed targets. The main part of this thesis is on signal processing of low frequency SAS in shallow water environments, including fast imaging algorithm for bistatic SAS, wide-angle SAS imaging algorithm, the combination of MIMO and SAS, and signal processing for SAS image enhancement.Synthetic aperture imaging algorithm is one of the most crucial and computationally intensive parts of SAS signal processing. The geometrical relation between the source, the target and the receiver of a bistatic SAS is far more complicated than that of a monostatic one, putting much challenge on the imaging algorithm. The conventional time-delay and sum (TDS) algorithm is quite time consuming and may not able to be carried out in real- time. Fast imaging algorithms more or less put some requirements on the movements of the array. Based on the available monostatic SAS imaging algorithms and satellite borne bistatic Synthetic Aperture Radar (SAR) imaging algorithms, this thesis develops a fast bistatic SAS imaging algorithm, which is not limited to a straight-line track of SAS. Simu-lation and experimental results have verified the effectiveness of the developed algorithm.Because of the symmetrical characteristic of the shape of man-made targets, TSP presents deep angular dependence within a wide observation angle. The fluctuation of echo intensity degrades the performance of conventional SAS imaging algorithm. By modeling TSP of some primitive shaped targets, this thesis developes a Generalized Likelihood Ratio Test based SAS imaging algorithm:GLRT-SAS, whose core idea is accurate match filtering in the along track dimension. Simulation and experimental results reveals that GLRT-SAS can estimate attitude of a target accurately, and discriminate between man-made targets and nature targets.The high resolution in the along track dimension comes from coherent processing of signal from different pings, which can be considered as phase matching. Uncompensated phase error of echo signal may result in shifting and widening of the main lobe, or raising of side lobes. This thesis discusses the effect of motion mismatch and time asynchroniza-tion on bistatic SAS, and develops countermeasures respectively. In shallow water environ-ments, multipath propagation arising from acoustic boundary interactions could degrades SAS imaging performance, leading to ghost targets and reduced image contrast. Based on the method of robust wideband adaptive beamforming and time-varying steering, this the-sis develops a signal processing algorithm for short arrays, in which multipath signals are utilized to enhance SAS images.One of the main characteristic of anisotropic TSP is the limited main lobe, which will results in a sound shadow region when the target is illuminated by a stationary source. This limits the aperture that can be synthesized, and in consequence the coherent gain that can be achieved. A MIMO sonar can achieve spatial diversity by separation of transmitters, but also achieve waveform diversity by transmiting different waveforms. This thesis constructs a framework of coherent and non-coherent processing, where the concept of MIMO and SAS are closely integrated to benefit the detection and localization of acoustic targets.During the thesis research, a prototype shallow-water low-frequency bistatic SAS with stationary transmitter and an Autonomous Underwater Vehicle-AUV platform has been designed and built, including sonar system hardware and real-time control and data ac-quisition software. As a typical small-size and inexpensive AUV, the size and accuracy of navigation equipments of the "Dolphin01" are limited. An integrated navigation method is developed based on the extended Kalman filter, in which two novel processes, track-based compass calibration and adaptive beamforming-based SAS micro-navigation, are incorporated. Through tank test and lake experiment, the prototype SAS system has shown working reliably while acquiring a large amount of real data. The processing results and analysis of the experiment data have verified the performance of relevant signal processing algorithms proposed in this thesis.