An algorithm for traffic grooming in WDM optical mesh networks with multiple objectives

Abstract

-This paper studies a traffic grooming in wavelength-division multiplexing (WDM) mesh networks for the SONET/SDH streams requested between node pairs. The traffic could be groomed at the access node before converting to an optical signal carried in the All-Optical network. We design a virtual topology with a given physical topology to satisfy multiple objectives and constraints. The grooming problem of a static demand is considered as an optimization problem. The algorithms found in the literatures focus on a single objective either to maximize the performance or to minimize the cost. We propose a Multi-Objective Evolutionary Algorithm to solve a grooming problem that optimizes multiple objectives all together at the same time. In this paper we consider the optimization of three objectives: maximize the traffic throughput, minimize the number of transceivers, and minimize the average propagation delay or average hop counts. The simulation results show that our approach is superior to an existing heuristic approaches for the 6-node benchmark network in an acceptable running time.KEYWORDS -Multi-Objective Evolutionary Algorithm, Grooming, WDM, mesh optical networks, Pareto Optimal. INTRODUCTIONA grooming is a process of multiplexing low-speed traffic onto the high-capacity pipe. It is a challenging technology that could cut costs, improve the protection and decrease provisioning time in the Optical network. While the Optical switch can handle up to OC-768, the electronic switching technologies at the access node can only feed the rates much lower than the full capacity of the Optical network. Therefore, the grooming is the right solution to lower this gap.Literatures on grooming are mainly based on a ring topology. Our research work is on the traffic grooming in WDM mesh networks with static traffic demand that is similar to [1]. It is considered as an optimization problem. The main purpose is to design the virtual topology that optimizes both a performance and a cost. The objectives include a maximization of a throughput as found in [1], a minimization of the network cost like number of transceivers as found in [2] and a minimization of the average propagation delay of the lightpaths. To achieve these objectives, we propose a multi-objective evolutionary algorithm to search for the solution.The structure of the remainder of this paper is as follows. Section II presents mathematical formulation of the virtual topology design in the traffic grooming problem. Section III explains the Multi-Objective Evolutionary Algorithm in general. Then in Section IV, we propose how to apply the MOEA for the traffic grooming problem. Section V shows the performance of our approach comparing to the heuristics on a 6-node network. The conclusion of our approach and suggestions for the future work are in Section VI.

Multiobjective swarm intelligence for the traffic grooming problem

“…Several grooming policies and traffic selection methods are also presented. These grooming policies are the following: Minimizing the Number of Traffic Hops In [14], the aim of the authors is to design a virtual topology that optimizes performance and cost. Their objective functions include a maximization of throughput (as found in [22,24], and [11]), a minimization of transceivers (as found in [10] and [8]), and a minimization of the average propagation delay of the lightpaths.…”

“…Then, we update the best and the worst value of MOFitness, and the Gravitational constant (G) (lines [13][14][15][16]. After that, we calculate the mass and acceleration of each agent of P (lines 17-33).…”

A fault tolerant channel allocation scheme in distributed cloud networks

“…Moreover, the bandwidth request of a DC traffic stream can be much lower than the capacity of the lightpath. Efficient grooming of DC's connections onto high-capacity lightpath will improve the network throughput and reduce network management cost [11,12].…”

Minimisation of end‐to‐end delay in reconfigurable WDM networks using genetic algorithms

“…One of these algorithms is minimum-delay logical topology design algorithm (MLDA), which aims at designing logical topologies with low end-to-end delay considering both the physical network and the traffic demand. In Reference [11], a multi-objective evolutionary algorithm is proposed by Prathombutr et al to simultaneously maximise the throughput and minimise the number of transceivers and the propagation delay, but it does not take into account the transmission and routing delays. Finally, Ghose et al [12] propose two algorithms (one heuristic and one genetic) to design logical topologies minimising the end-to-end delay, and they also solve the RWA problem.…”