Multiple-Layer Network Planning with Scenario-Based Traffic Forecast

Abstract

One of the major tendencies in network planning is to take the interactions of multiple network layers into consideration, so that solutions much better than those restricted in single layers can be achieved. This also applies when planning for the extension of the network infrastructure to meet the increment of traffic in the future, in which minimizing the capital expenditure of new devices and the corresponding energy consumptions are among the greatest concerns of network operators. However, the planning relies on the forecast of future traffic demands, where some degree of uncertainty must be counted as a part. A special forecast mechanism currently used by some network operators is to make a series of forecasts, each being a scenario which describes a possible future situation. In this paper, we will introduce this mechanism, analyze its differences to other traffic uncertainty models, and suggest the corresponding planning methods based on the Integer Linear Programming (ILP) models. The performances are shown in the tests at a scale of real problems.