A comparison of multi-objective optimization algorithms for weight setting problems in traffic engineering
- PDF / 633,475 Bytes
- 16 Pages / 595.276 x 790.866 pts Page_size
- 62 Downloads / 186 Views
A comparison of multi-objective optimization algorithms for weight setting problems in traffic engineering Vítor Pereira1
· Pedro Sousa2 · Miguel Rocha1
Accepted: 3 September 2020 © Springer Nature B.V. 2020
Abstract Traffic engineering approaches are increasingly important in network management to allow an optimized configuration and resource allocation. In link-state routing, setting appropriate weights to the links is an important and challenging optimization task. Different approaches have been put forward towards this aim, including evolutionary algorithms (EAs). This work addresses the evaluation of a single and two multi-objective EAs, in two tasks related to weight setting optimization towards optimal intra-domain routing, knowing the network topology and aggregated traffic demands and seeking to minimize network congestion. In both tasks, the optimization considers scenarios where there is a dynamic alteration in the network, with (1) changes in the traffic demand matrices, and (2) link failures. The methods will simultaneously optimize for both conditions, the normal and the altered one, following a preventive TE approach. Since this leads to a bi-objective function, the use of multi-objective EAs, such as SPEA2 and NSGA-II, came naturally; those are compared to a single-objective EA previously proposed by the authors. The results show a remarkable performance and scalability of NSGA-II in the proposed tasks presenting itself as the most promising option for TE. Keywords Traffic engineering · Intra-domain routing · Link-state routing protocols · Multiobjective evolutionary algorithms · NSGA-II
1 Introduction Congestion avoidance in whole or part of an Internet Protocol (IP) network is one of the most important problems for Internet Traffic Engineering (TE). Distinct solutions that target optimal traffic congestion levels on a network have emerged in the networking research community, resorting to diversified strategies. Some of those proposals are reactive, and try to avoid congestion by adapting traffic flows at the edge of the network (Villamizar and Song 1994; Floyd and Fall 1999). Other solutions, enabled by new trends like Software Defined
B
Vítor Pereira [email protected] Pedro Sousa [email protected] Miguel Rocha [email protected]
1
Department of Informatics, Centre of Biological Engineering, University of Minho, Braga, Portugal
2
Department of Informatics, Centro Algoritmi, University of Minho, Braga, Portugal
Networking (SDN), maximize the network utilization with hybrid deployments, where some flows are directed according to routing protocol decisions, while others are forwarded reflecting specific administratively installed rules (Jain et al. 2013). There are also preemptive approaches that consider known or estimated traffic requirements, and try to avoid congestion by optimizing traffic distribution on the available resources (Fortz 2000). Regardless of the congestion avoidance strategy, at the core of the problem lies the necessity to improve resources management and, in thi
Data Loading...
