Hopfield Network Approach to Beamforrning in Spread Spectrum Communication

This paper discusses a neurobeamformer based on the Hopfield neural network which is used to suppress narrowband interference in spread spectrum communications. In this application, we considered a liiear array consisting of four spatially separated elements. The received spread spectrum signal is bi-phase modulated by a PN (Pseudo-Noise) code. This code has parameter which will prove to be well suited for use in network Comparators to obtain optimal antenna array pattem, i.e., the main beam is steered towards the desired signal while nulls are directed towards the interference. The proposed Hopfield beamformer can be characterized as a constrained quadratic function. It uses random, asynchronous updates in order to provide real time response to rapid time-varying environments. The constrained quadratic programmed neural network has an associate energy function which the network always seeks to minimize, which leads to optimization of the array weights. In this paper we present simulations carried out with a Hopfield beamformer. It will be shown that its performance is better than that of a LMS beamformer.