Simple Neural Model for Approximate Localization of Mobile Stochastic EM Sources with Variable Signal Level and Linear Trajectory in The Azimuthal Plane

Abstract

In this paper, a simple neural model is presented for approximate determination of the angular position of stochastic radiation sources that move along the linear trajectory at azimuthal plane and whose incoming signal level fluctuates with the change of their position. The neural model is based on a multilayer perceptron network (MLP). In order to perform the localization of stochastic sources, the signals of these sources are sampled with a linear uniform antenna array, and then based on the output of the antenna sensor, spatial correlation matrix is formed, and at the end, the values of this matrix are input to the MLP network determining the angular positions of the source.