Neural Network Based Spread Spectrum PN Code Acquisition System

In this paper, a neural network based direct sequence spread spectrum code synchronization system is proposed. This system is based on training a recurrent random neural network (RNN) model on all the possible phases of the used spreading code. The trained network can then be used at the receiver for the initial coarse alignment of the local code phase and the received code. One advantage of this technique over the conventional synchronization techniques is that the phase of the received PN code can be decided without searching the potential code phases. Also the RNN, after being trained, can have a simple hardware realization that makes it candidate for implementation as a dedicated chip. This makes the neural network based technique faster and more robust than the conventional techniques. Computer simulations, carried out on maximal length sequences of length N = 7 and N = 15, show that the proposed system can e ectively indicate the phase of the received code even with very low signal to noise ratios.