Neural networks based signal detection

The aim of this paper is to present different neural networks able to realize a radar detector. Multilayer perceptrons are considered with different structures which make use of different backpropagation algorithms during the learning phase of the neural network. The reference detection scheme assumed for comparison purposes is a coherent integrator followed by an amplitude detector and optimum thresholding. The comparison (in the Neymann-Pearson sense) with the optimum detector performances allows to assess the signal-to-noise losses pertaining to the different neural network detector structures.<
>