Design and analysis of sparse MIMO Array and sparse recovery algorithms for super-resolution DOA estimation

More Info
expand_more

Abstract

The development of Smart Vehicles (SV) has increased the demand for secure and intelligent sensors. The automotive radar plays a massive role in improving the security of these vehicles. Radar needs to make fast and accurate detection in a noisy environment while being aware of its surroundings. The modern radar systems deployed on the SVs utilize multiple sensors to keep track of their surroundings and improve radar cognition. Even though adding more sensors will help make accurate decisions, the processing time to make those decisions may be affected. Hence, the research focuses on improving the accuracy of the decisions without adding extra sensors and extra processing time. The recent development of the Compressed Sensing (CS) theory has provided new techniques to reduce the number of measurements required for storing the signal and recovering the signal. This idea can be used for Direction-of-Arrival (DOA) Estimation, where we have very few measurements to estimate accurately. Sparse recovery algorithms based on the CS theory have shown promising results for single snapshot DOA estimation. Uniform Linear Array (ULA) provides redundant spatial frequency samples. This redundancy can be reduced by removing specific elements from the array. Removing the redundant elements can help improve the radar's aperture size and angular resolution; these arrays are known as sparse arrays. Combining sparse recovery algorithms with sparse arrays, the angular resolution and accuracy of the DOA estimates can be improved. Based on this idea, an optimal array search algorithm has been proposed in this thesis. The design technique optimizes the Multi-Input-Multi-Output (MIMO) array configuration for improving sparse recovery guarantee. Optimal MIMO topologies, as an example for 2Tx4Rx and 3Tx4Rx (Tx-Transmitters, Rx-Receivers), have been synthesized. The performance of these arrays has been tested with prominent sparse recovery algorithms. The performance of the algorithms is also ranked based on their probability of detection and angular resolution. Improvement in the angular resolution up to 8 degrees with respect to the ULA-MIMO for 2Tx4Rx configuration and up to 5 degrees for 3Tx4Rx configuration is obtained with the help of a sparse recovery algorithm.