Coherent Signal Direction Finding With Sensor Array Based on Back Propagation Neural Network

Abstract

An algorithm based on back propagation neural network and particle swarm optimization is proposed to solve the direction of arrival (DOA) estimation of coherent signals received by the sensor array in colored noise environment. First, a spatial differential smoothing algorithm is adopted to eliminate colored noise and the independent signals to obtain a covariance matrix only containing the coherent sources. Then, the first line of the covariance matrix is extracted as an input characteristic parameter vector, meanwhile, the DOA of the coherent signals are taken as output. Finally, the trained back propagation neural network optimized by particle swarm algorithm is exploited to reckon the directions of coherent signals. The algorithm put forward in this paper does not require eigen-decomposition and spectral peak searching, so the computational burden is low. Theoretical analysis and simulations demonstrate that the proposed algorithm has high angular resolution and direction finding accuracy in colored noise environment.INDEX TERMS Coherent signals, back propagation neural network, colored noise, particle swarm optimization.

Heating and Cooling Loads Forecasting for Residential Buildings Based on Hybrid Machine Learning Applications: A Comprehensive Review and Comparative Analysis

“…Input signals with weight are received by the input layer and transmitted to the hidden layer after processing. As an internal processing layer, the hidden layer processes the signals and transmits them to the output layer [85]. The neural network training process requires an activation function; 𝑓(𝑋) is the sigmoid activation function used in this paper [79]:…”

Novel Approach to 2D DOA Estimation for Uniform Circular Arrays Using Convolutional Neural Networks

“…M eigenvalues can be obtained by eigenvalue decomposition of the UCA covariance matrix of M elements. In order to attain the input characteristic N l − R xx of the 2D DOA neural network, i.e., equation (12), at least one eigenvalue corresponding to the noise power must be guaranteed. erefore, the maximum number of targets that can be estimated by the proposed method is M − 1.…”

“…In recent years, neural network-based methods have been extensively developed for improving the operation speed and adaptability of DOA estimation. In order to achieve 1D DOA estimation, references [12][13][14][15] employ a uniform linear array (ULA). e former use the multilayer perceptron (MLP), and the latter support vector machine (SVM).…”

“…e MLP, SVM, and RBF methods require the input feature of neural networks to be a vector, which might cause the input feature dimension to be excessively high. To avoid the problem, references [12,[17][18][19] adopt the strategy of taking only the first row of the preprocessed array covariance matrix as the input feature vector while ignoring noise interference on each element of the covariance matrix. erefore, reference [14] adopts the strategy of averaging diagonal elements of the preprocessed array covariance matrix to slightly reduce noise interference.…”