Multi-Period Planning of WDM Transport Networks Incorporating Demand Uncertainties

In this work, we study the dimensioning of an optical network over multi-period planning horizons, incorporating the uncertainties of the demand forecast. Specifically, we consider the problem of adding capacity incrementally on a point-to-point wavelength-division multiplexing WDM system over a finite time horizon, so as to meet the evolving demands while minimizing the total effective system cost (installation cost and expected penalty cost due to capacity shortage) incurred. To gain analytical insights, we setup this capacity expansion problem in the framework of inventory control theory, and we solve the problem using dynamic programming technique. The optimization results enable us: 1) to gain quantitative insights on the impact of long-term demand variability and technological advance on the network design; 2) to analyze the trade-off between the policy of installing capacity earlier ("pay now") and the policy of delaying capacity installation later ("pay as you grow"); 3) to evaluate the merits of using low-density or high-density line systems