Scalable Multi-objective Optimization of Reliable Latency-constrained Optical Transport Networks

Optimization over time of reliable 5G-RAN with network function migrations

“…This work extends [9] in several aspects. A fast and effective heuristic addressing the computational bottlenecks in the proposed methodology is developed.…”

“…DUOpt splits the large multi-objective Integer Linear Programming (ILP) problem into several smaller single-objective ILP subproblems, which are solved sequentially according to their application-defined priority ordering. The bottlenecks in the optimization process are discussed, and a hybrid approach, partially based on local search, is described, which extends and improves previous contributions [9]. An extensive comparison with a classical aggregate multi-objective optimization approach is presented both in regular lattice networks and nonregular ones, showing the effectiveness of the lexicographic algorithm in solving the multi-objective problem in short computing times.…”

“…13, at 8:00 there are 5 new active nodes with respect to 7:30, for a total of 10 active nodes. Five new nodes (3,9,25,32,38) have been activated; with the previous solution fixed, that would have led to 11 active nodes. In fact, since the objective function of DUMig rewards the deactivation of nodes that are no longer needed, node 2 has been turned off.…”