A Novel Integer Linear Programming Formulation for Designing Transparent WDM Optical Core Networks

Abstract

The explosive growth of Internet traffic driven by the rise and popularity of bandwidth-intensive services such as virtual realities, big data exchanges and artificial intelligence necessitates for a re-consideration of optical transport networks toward greater capital and operational efficiency from both designing, planning and management perspectives. Thanks to mainly the significant advances in transmission technologies and reconfigurable optical add/drop multiplexers, transparent (all-optical) core networks with wavelength-division multiplexing (WDM) technologies have been well-adopted by network operators as costly optical-electrical-optical interfaces are eliminated together with the merit of scalable upgrade. Efficient designing of such transparent networks therefore has been of ever-growing importance as the spectrum capacity has to be optimized while maintaining the efficiency of operations. Conventional approaches based on integer linear programming formulations and/or heuristic algorithms have nevertheless been focused on optimizing a single metric, i.e., wavelength count or wavelength link usage, and doing so possibly poses penalties to the remaining objectives. In addressing this issue, this paper presents a new integer linear programming formulation incorporating both two performance metrics into consideration with an integrated objective function. We then introduce a framework for analyzing and properly setting up the weight coefficients of integrated objective function so as to guarantee the order of optimization. Extensive numerical studies on realistic topologies highlight the effectiveness of our proposal compared to conventional designs based on single objective formulation in the literature.