A Genetic Algorithm based solution to three dimensional Direction of Arrival (DOA) problem

Abstract

Estimating Direction of Arrival (DOA) of a signal in three dimensional space is a procedure which is difficult to perform in a reverberant and noisy environment, yet it demands high degree of precision. One of the most reliable techniques under reverberant conditions is Multi-Sensor Steered Response Power with Phase Transform filter (SRP-PHAT) but it is computationally exhaustive if performed using traditional calculus based techniques. So this paper proposes a different and computationally efficient approach to solve the DOA problem, one that is based on searching and optimization capabilities of Genetic Algorithms. In this paper, for the purpose of application of genetic algorithm, SRP-PHAT to estimate DOA is modeled as an optimization problem. To speed up the process of global convergence of genetic algorithm, this paper describes a procedure which can act as a precursor for the design and implementation of a parallel architecture for DOA estimation. To prove the efficacy of the solution, various performance curves have been included which show that the proposed solution is superior to traditional techniques and the last section describes the computational load of the proposed technique and compares it with two other conventionally employed techniques.