A DRA scheme based on Hopfield neural networks methodology

Abstract

This paper proposes a dynamic resource allocation (DRA) algorithm that makes use of the Hopfield neural network (HNN) methodology, which provides a fast way of finding the optimum resource when formulated as a combinational problem. The proposed algorithm is applied to both the scheduling of downlink transmissions in a CDMA scenario with delay-oriented services, and real time services, although by a proper modification of the constraints imposed in the energy function, it could be easily extended to other services or access technologies. The algorithm is evaluated by means of simulations, revealing its ability to adapt to the specific service and traffic conditions. Finally, the convergence issues related to HNN methodology are considered in the light of the obtained results