Solving large telecommunications network loading problems

Telecommunications network loading problems focus on routing traffic through networks so as to satisfy the demand forecast between all pairs of nodes. The objective is to minimize the total routing costs subject to satisfying capacity constraints associated with the individual links. This paper presents a nonlinear programming formulation for the multiperiod network loading problem, a problem that appears in many guises throughout network planning. Such problems are also known in the operations research literature as (multiperiod) multicommodity network flow problems. The resulting optimization problem has a very large number of variables and constraints. We describe the implementation of a nonlinear optimization algorithm, the Frank-Wolfe method with PARTAN extension and projective gradient acceleration, on a parallel vector processor for solving these large dynamic network loading problems. Our computational experience shows that our algorithm solves large, realistic loading problems very efficiently and is serving as a valuable tool for telecommunications network planning.