Neural Network Methods with Traffic Descriptor Compression for Call Admission Control

We present and evaluate new techniques for call admission control based on neural networks. The methods are applicable to very general models that allow heterogeneous traac sources and nite buuers. A feedforward neural network (NN) is used to predict whether or not accepting a requested new call would result in a feasible aggregate stream, i.e., one that sat-isses the QOS requirements. The NN input vector is a traac descriptor for the aggregate stream that has the following beneecial properties: its dimension is independent of the number of traac classes; and it is additive, allowing it to be updated eeciently by simply adding the traac descriptor of the new call. A novel asymmetric error function for the NN helps achieve our asymmetric objective in which rejecting an infeasi-ble stream is more important than accepting a feasible one. We present an NN design that provides an optimal linear compression of the NN inputs to a smaller number of traac parameters. The special case of one compressed parameter corresponds to an NN version of equivalent bandwidth. Experiments show our methods to be better than methods based on equivalent band-width, with respect to call blocking probability, through-put, and the percentage of feasible streams that are correctly classiied.