Compressive Sensing Based Estimation of Direction of Arrival in Antenna Arrays

This thesis is concerned with the development of new compressive sensing (CS) techniques both in element space and beamspace for estimating the direction of arrival of various types of sources, including moving sources as well as fluctuating sources, using one-dimensional antenna arrays. The problem of estimating the angle of arrival of a plane electromagnetic wave is referred to as the direction of arrival (DOA) estimation problem. Such algorithms for estimating DOA in antenna arrays are often used in wireless communication network to increase their capacity and throughput. DOA techniques can be used to design and adapt the directivity of the array antennas. For example, an antenna array can be designed to detect a number of incoming signals and accept signals from certain directions only, while rejecting signals that are declared as interference. This spatio-temporal estimation and filtering capability can be exploited for multiplexing co-channel users and rejecting harmful co-channel interference that may occur because of jamming or multipath effects. In this study, three CS-based DOA estimation methods are proposed, one in the element space (ES), and the other two in the beamspace (BS). The proposed techniques do not require a priori knowledge of the number of sources to be estimated. Further, all these techniques are capable of handling both non-fluctuating and fluctuating source signals as well as moving signals. The virtual array concept is utilized in order to be able to identify more number of sources than the number of the sensors used. In element space, an extended version of the least absolute shrinkage and selection operator (LASSO) algorithm, the adaptable LASSO (A-LASSO), is presented. A-LASSO is utilized to solve the DOA problem in compressive sensing framework. It is shown through extensive simulations that the proposed algorithm outperforms the classical DOA estimation techniques as well as LASSO using a small number of snapshots. Furthermore, it is able to estimate coherent as well as spatially-close sources. This technique is then extended to the case of DOA estimation of the sources in unknown noise fields. In beamspace, two compressive sensing techniques are proposed for DOA estimation, one in full beamspace and the other in multiple beam beamspace. Both these techniques are able to estimate correlated source signals as well as spatially-close sources using a small number of snapshots. Furthermore, it is shown that the computational complexity of the two beamspace-based techniques is much less than that of the element-space based technique. It is shown through simulations that the performance of the DOA estimation techniques in multiple beam beamspace is superior to that of the other two techniques proposed in this thesis, in addition to having the lowest computational complexity. Finally, the feasibility for real-time implementation of the proposed CS-based DOA estimation techniques, both in the element-space and the beamspace, is examined. It is shown that the execution time of the proposed algorithms on Raspberry Pi board are compatible for real-time implementation