Neural networks in communications: Admission control and switch control

Abstract

A review is presented of recent work on neural networks, particularly two models which have found applications to communication network resource management. The authors describe results from using neural networks for call admission control and switch control. The call control problem is the selective admission of a set of calls from a number of inhomogeneous call classes, which may have widely differing characteristics. The approach adopted is that key network performance parameters are observed while carrying various combinations of calls, and their relationship is learned by a neural network structure. The neural network model chosen has the ability to learn from past results and to adapt to changing conditions. The switch control problem is the service policy used by a switch controller in transmitting messages. In a crossbar switch with input queuing, significant loss of throughput can occur when head-of-line service is employed. Various alternatives to first in, first out (FIFO) input queuing are examined and their performances are compared.