Otimização multi-objetivo aplicada à alocação dinâmica de rotas em redes de telecomunicações

Abstract

The new networking technologies and the Internet difusion produced the data traffic increase in the networks. The quality of service (QoS) requirements for applications also increased and become stricter. Different QoS metrics can be evaluated by applications, for example the delay, the load balance and the packets lost in the communication.Different optimization algorithms can be used to achieve the QoS metric values for applications. The single objective approach regards as objective function one QoS metric and the others are considered as problem constraints, while the multi-objective optimization considers each QoS metric as an objective function and optimize them together. Because different objective function are optimized at once, the multi-objective result consist in set of solutions, named Pareto-optimal set.To apply multi-objective optimization for dynamic routing problems is a great challenge, because this strategy demands a decision support system to choose an appropriated solution from Pareto-optimal set dynamically, matching the QoS requirements. There is no works investigating this strategy in the literature, and this is the contribution proposed in this paper.To verify the performance of this strategy in telecommunications networks context, the MPLS architecture (Multi-Protocol Label Switching) is used . This architecture extends the functionalities of the IP protocol and presents as main advantage the possibility to perform explicit routing of requests in the network, defining explicitly the path to data traffic flow for each communication request.The results show improvements in overall performance of the network evaluating different QoS metrics simultaneously. Another advantage is the flexibility provide by the multi-objective optimization allowing to choose different solutions for the problem by changing the parameters in the decision support system.