Neural Network Application to Telemetry Frame Synchronization

Abstract

This paper looks into the use of neural network software as applied to the classical signal to noise concern when dealing with space to ground data communications. Use of a digital neural network to extend the correlation range of Pulse Code Modulation (PCM) down into noise is investigated. Conventional synchronization pattern correlation is done with digital logic comparisons on a sliding window with a set number of bit mismatch errors allowed. Correlation with a neural network does pattern recognition with a weighted network of artificial neurons that have been trained to recognize the sync pattern within noise. The output of such a neural network will produce a best guess of the correct pattern.