An adaptive traffic prediction algorithm for cellular radio systems

Abstract

Prediction of customer demand is an important aspect of the design of any communication system. Accurate predictions enable the system operator to allocate resources efficiently such that the quality of service offered is maximised whilst the capital expenditure on infrastructure is minimised. Within the paper an automated technique for the prediction of customer demand, or system traffic, within a cellular radio system is presented. Within the technique a model for the traffic generation process is firstly assumed. The parameters of the model are then derived using an adaptive algorithm to minimise the error apparent between the prediction obtained from the model and actual traffic data available from the real system. Once the model parameters have been derived it is then possible to predict traffic in areas in which actual data is unavailable or in situations in which system parameters may be subject to alteration. >