Performance Evaluation of Direction of Arrival Estimation Using Uniform and Non-Uniform Linear Arrays for Signal Source Localization

Abstract

Due to the flexibility and convenience of wireless communication, many applications began to adapt to it. Hence, to comply with the requirements, a high speed data rate is necessary. With the advancement of technologies and plethora of applications, wireless communication tend to be overloaded which has resulted in the utilization of the higher frequencies in the spectrum owing to the larger bandwidth attributed to the higher band. The higher data rate, and consequently multipath fading and interference result in the limitation of data rate transmission. Therefore, the objective of this research project is to determine the desired signal location, minimize the multipath and interference by analyzing subspace techniques for direction of arrival (DOA) estimation using uniform linear array (ULA) and non-uniform linear array (NLA). Especially, this thesis focuses on enhancing the wireless communication capacity by estimating the DOA using Multiple Signal Classification (MUSIC) and Root-MUSIC algorithms. To apply these algorithms on the ULA and the NLA helps to analyze the accuracy and efficiency DOA estimation. The first phase of this thesis is an extensive study of various high resolution directions of arrival estimation algorithms such as MUSIC and Root-MUSIC algorithms. Then apply them on ULA. The second phase is to estimate DOA using NLA. Thereafter, decompose it into two ULAs for the corresponding co-prime array before combining the two ULAs with MUSIC algorithm. The simulation of the DOA estimation with variation of snapshots, signal to noise ratio, array elements, signal source as a set of input parameters is carried out in MATLAB platform. Their performance under ULA and NLA are tested and the evaluation analysis of their accuracy and resolution towards the direction of arrival (DOA) estimation is checked. The analysis is based on the evaluation of the DOA estimation performance using ULA and NLA in the presence of additive white Gaussian noise and the calculation of the estimation subspace methods through MATLAB platform. The performance of the algorithms has been analyzed by considering Mean Square Error (MSE) for eight trials as a function of array elements, of signal to noise ratio and as function of snapshots. Through extensive simulations NLA has shown to be more accurate and efficient in DOA estimation.