A method is proposed for deriving a traffic characteristics model that can be used to forecast the traffic volume for intelligent telecommunication services. A sort of regression analysis with dummy variables is used to represent the service quantitatively and to construct the traffic characteristics model. Recursive least squares estimation, which is a special case of the Kalman filter, is applied to the traffic characteristics model to forecast the traffic volume. In the proposed modeling and forecasting, qualitative factors representing a certain service attribute are selected and using an information criterion, the model with the best fit is identified as the most suitable forecasting model. Numerical results using practical observation data showed that the proposed method produces an accurate forecast and is thus effective for practical use.

In an ATM network, there are quality impairments particular to the ATM network such as cell loss and delay variation. During ATM network planning, therefore, various causes of quality impairments should be clarified. This paper overviews ATM network performance issues, and discusses performance requirements for the SDH network which will be applied as a physical layer of the ATM network. It also presents ATM network performance planning methods on cell loss and cell delay.