Airborne Radar Ground Clutter Suppression Using Multitaper Spectrum Estimation & Choosing DPSS Parameters

Sammanfattning: One of the biggest challenges in any airborne radar is to distinguish a target from a strong ground echo. The main problem is that the ground echo, called ground clutter, can be up to a million times stronger than the response from the target in question. Today many different filtering methods are used in airborne radar systems to separate the target signal from the ground clutter. All of them with their own advantages and shortcomings. In an ideal world the optimum filter would completely filter out the unwanted ground echo. But as ideal filters don't exist in reality a filter with low sidelobes and minimum loss in signal-to-interference ratio is sought after. A type of filter which exhibit this behaviour are discrete prolate spheroidal sequences (DPSS). This thesis investigated if DPSS could be used as weight functions in multitaper spectrum estimation to filter out ground clutter in the radar signal. A simple clutter model was developed for generating simulated ground clutter which was then filtered out by multitaper and a traditional method. Results showed that it is possible to use DPSS in multitaper spectrum estimation and that it outperforms a basic traditional method in clutter filtration as long as parameters such as bandwidth and the number of sequences used are chosen properly. The increase in performance against the traditional method comes at a cost of increased computational load with each additional DPSS order used. A full factorial experiment was also performed to investigate which parameters were important for maximising improvement factor and minimum detectable velocity. The results from these showed that a low bandwidth in the generation of the DPSS was preferable and that a high number of time samples and DPSS used improved performance. They also showed that for an increase in number of time samples the bandwidth and number of sequences used need to be adjusted to maintain the same level of the improvement factor. It was concluded that future work should focus on validation with more advanced clutter models and MTI filters in simulations as well as validation against real radar data. If proved successful, optimisation of calculation speeds as well as implementation of adaptive choice of DPSS bandwidth would be beneficial before being implemented in a radar system.