Access network dimensioning with uncertain traffic forecasts

Network planning remains a recurrent and critical task for telecommunication operators. One of the difficulties is to deal with traffic forecasts. Indeed, these data are often not reliable, in the sense that they do not fit well the occurring reality. The aim of the present paper is to provide mathematical models for access network planning, which take account of forecast uncertainty. Taking advantage of classical stochastic programming models, equations are derived to characterize a network dimensioning able to accommodate the traffic with a targeted probability. Several probabilistic frameworks are proposed. Finally, preliminary numerical results are given on a realistic instance. They illustrate the possible benefit of the proposed approach.