A High‐Resolution DOA Estimation Method by Compressive‐Sensing Based Sparse Minimum Redundancy Linear Array for IoT Sensory Networks

ABSTRACT IoT sensors are increasingly being utilized for direction of arrival (DOA) estimation, which is crucial for tracking signal sources and enhancing localization in various applications, particularly in wireless sensor networks. The ability of IoT sensors to retrieve direction information from the received electromagnetic waves enables them to effectively function in heterogeneous environments, thus facilitating more accurate monitoring and localization of sources. In recent years, linear antenna arrays, with spatially sparse positioning of elements has evolved as one of the most significant arrangement of arrays, in IoT sensory networks intended for source localization or DOA estimation of targets in the far field region. Sparse arrays yield with a higher degree of freedom (DOF) and can produce a larger antenna aperture as compared to uniform linear arrays (ULA). Various kinds of linear sparse arrays have been familiarized and researched in the last decade, such as co‐prime array, nested array, super nested array, etc. The minimum redundancy linear array (MRLA) structure is a class of sparse arrays introduced few decades ago and is essentially a linear array arrangement, which reduces the redundant spacings between the array elements, to attain maximum resolution. Along with the investigation of sparse and nonuniform array geometries, a new paradigm of sparse signal reconstruction of underdetermined system of linear equations became apparent in the last decade, known as compressive sensing methodology. Quite a few benefits have transpired with compressive sensing technique in array signal processing, including a significantly reduced requirement of sensory devices, a substantial reduction in memory storage, a higher data transfer rate, and a considerable reduction in computational complexity. In this paper, an enhanced DOA estimation technique is proposed by integrating the MRLA sensory structure with the compressive sensing framework for single snapshot instance. The MRLA model's low‐dimensional kernel compresses incoming signals to preserve the large array aperture. The high‐resolution DOA estimation method is subsequently executed using these compressed signals through a proposed modified norm‐based reconstruction process. Several simulation experiments with various parameters are performed and compared with the popular standard models to verify the performance and ascendancy of the proposed method, such as computational complexity, failure rate, root mean square error (RMSE) of the estimated DOA, probability of resolution, with variation of signal‐to‐noise ratio (SNR) values, etc. The comparative results indicate that the proposed method outperforms standard techniques in more accurate DOA estimation, specifically at lower SNR values.