Dynamic channel allocation based on a Hopfield neural network and requirements for autonomous operation in a distributed environment

This paper presents a distributed architecture that provides a platform for performance study of cellular radio dynamic channel allocation (DCA) schemes based on a Hopfield neural network (HNN). HNN-based DCA schemes have been studied over the recent years but these techniques are invariably centralised. Operation in a distributed environment necessitates certain system level requirements for accessing system-wide channel allocation information used by the algorithm. Distributed operation also suggests that the HNN-based DCA schemes operate autonomously and it would be of interest to evaluate how such schemes compare with the centralised schemes. The paper describes a suitable network architecture for a distributed HNN and highlights some of its key benefits through simulation