Connection rerouting in GRWA networks

Abstract

Traffic grooming consists of packing low rate streams onto a high speed lightpath in order to effectively use the network resources. Under dynamic traffic, rerouting of ongoing connections has been envisioned as a means, to be used very wisely, to reduce the connection blocking rate and to optimize the network resources. In a context of traffic with QoS constraints, only the delay tolerant ongoing connections are rerouted. In this paper, we design three new heuristics, one relying on a mathematical ILP (Integer Linear Program) model and two low complexity ones to carefully reroute ongoing connection requests in order to accommodate the new incoming connection requests, while minimizing connection disturbance. While the ILP model allows the full exploration of a limit on the overall number of reroutings, both heuristics are designed as low complexity heuristics in order to limit the number of rerouting per establishment of a new incoming connection request. Comparative computational results show that the two low complexity heuristics provide much better results in terms of the best compromise between maximizing the throughput and minimizing the number of disturbances of the already established connection requests.