A hybrid approach involving artificial neural network and ant colony optimization for direction of arrival estimation

Abstract

This paper discusses the application of a multi-layer perceptron network to estimate direction of arrival (DOA) using ant colony optimization (ACO) for training. ACO simulates the foraging behavior of ant colonies which manage to find the shortest path from nest to feeding source. This technique was originally developed for discrete optimization problems, but recent research efforts has led to some algorithm modifications to make it applicable to continuous optimization problems. In this work we utilize continuous ACO to train a neural network for direction of arrival estimation which encounters an interpolation of a complex nonlinear function. The performance of proposed hybrid approach is compared to radial basis function network that is a well known solution to DOA problem and some improvements in approximation are discussed.Index Terms-Direction of arrival, ant colony optimization, radial basis function neural network

A Review on Machine Learning-Based Radio Direction Finding

“…ey also considered using the DOAs as the NN output directly [32,33]. Similar inputs are considered in [34,35], except that Pour et al [34] used a MLP network along with ant colony optimization for NN training and a BP NN was used in [35]. A similar treatment for a uniform linear array was adopted in [36].…”

A Survey on Applications of Multi Layer Perceptron Neural Networks in DOA Estimation for Smart Antennas

“…The abilities of these methods to handle with correlated signals are limited. The MLE method is very accurate and can handle partially correlated signals but it is computationally intensive [32] [37]. The matrix pencil (MP) method needs a single snapshot thus requires less computational time [40].…”

“…The genetic algorithm (GA) helps the MLP to avoid getting stuck in local minima and achieve results with a smaller hidden layer (20 neurons in the hidden layer) as compared to the RBF in the study. In [37], use of ant colony optimization (ACO) method to train MLP having 5 neurons in the hidden layer has been used for DOA estimation. In both cases results have shown the effectiveness of these methods.…”

Amplitude and Directional of Arrival Estimation: Comparison Between Different Techniques

“…In this, the strength of Hopfield neural networks (HNN) has been exploited for the DOA estimation. In [14] hybrid approach, i.e., the neural networks and ant-colony optimization has been used for estimating DOA of sources impinging on linear array. In [15] GA has been used for 2-D DOA estimation based on maximum likelihood technique for uniform circular array.…”