Array signal processing with interconnected Neuron-like elements

Abstract

Estimation of angles of arrival of plane waves f r o m data observed at an array of sensors is performed with a network of interconnected, instantaneous, s a t u r a t i n g non-linear elements called neurons. T h e networks use the observed data to decide which among a large number of hypothesized angles of arrivals best fits t h e data. A possible stochastic-digital implementation of such a network is also indicated.

Direction finding networks based on the approximate maximum likelihood and covariance fit formulations

“…Consider first a discretazation Expanding (8) and identifying terms with (2) gives us the following connection weights and inputs to the network…”

“…Several approaches to the direction finding problem using feedback networks have been proposed in [2] and [3]. In this paper, we show that if there is apriori knowledge about the number of sources, a Conditional Maximum Likelihood solution with a feedback network realization can be obtained.…”

Estimating two-dimensional angles of arrival by neural networks

“…These methods, pioneered by Rastoqi et a1 [3] , could have significant benefits over the classical algorithms for AOA estimation. Possessing the collective computational ability, these networks are of the fast convergence.…”

“…In this paper, we extend the work in [3] to estimate the 2-D angles of arrival of multiple narrow-band sources from the sampled data of a rectangular sensor array. The proposed 2-D neural estimator include two layer network, .in which the first layer performs two 1-D spatial searches using the rows and columns of the rectangular array, and the second carry out a 2-D verification of the data obtained in the first layer 1 -D proccssing.…”

“…The formulation we present here is an extension of the work in [3]. To avoid performing a 2-D search, two onedimensional (1-D) neural estimator are…”