20-01-2016, 03:02 PM
ABSTRACT
This paper presents an intelligent approach for modelling Routing and wavelength assignment (RWA) problem in wavelength-routed Dense-Wavelength-Division-Multiplexing (DWDM) optical networks. A new idea based on Artificial Bee Colony (ABC) algorithm is introduced for solving RWA problem which is known to be an NP-hard problem. In the proposed ABC-RWA approach every food source represents a possible and feasible candidate lightpath between each original and destination node pair in demand matrix. The positions of food sources are modified by some artificial bees in the population where the aim is to discover the places of food sources. The food source with the highest nectar value seems to be a solution which is evaluated by the fitness function. The proposed approach is evaluated for both path length (propagation delay) and hops count optimization schemes for PAN EUROPEAN and NSFNET test bench optical networks. The performance of ABC based approach is compared with the Genetic Algorithm (GA) model for solving RWA problem under random and heavy load traffic models. Simulation results demonstrate the ability and efficiency of proposed ABC model for solving RWA in real-world optical networks. Furthermore a [/align]comparison study approves that ABC is faster than GA to hit RWA global optimization solutions due to less complexity and computational processing.
Introduction
Today’s many worldwide networking and communications systems and applications use high-speed optical transport networks as appropriate backbones for connecting buildings, cities and countries such as PAN EUROPEAN, NSFNET and COST optical networks [1]. Modern optical transport networks are supported by some modern software and hardware technologies such as optical fibers, optical cross connect (OXCs) and optical add drop multiplexers (OADMs) nodal switches and DWDM techniques which enable them to offer multi terabit per second transmission rates to customers and applications [2]. Dense wavelength division multiplexing [3] technology divides the bandwidth of a typical optical fiber into some non-overlapping channels where each of them can operate at a different wavelength and,therefore, provides the opportunity of exploring the tremendous bandwidth of fibres in optical networks. In general, the DWDMoptical networks are considered as connection-oriented networks which mean that connections are established between node pairs before data transmission. Connection establishment in DWDM networks involves two main operations: routing and wavelength assignment [4,5]. In wavelength-routed DWDM optical networks the RWA problem [6,7] is a key designing and planning issue which refers to finding optimal routes for connection requests in demand matrix and assigning wavelengths to them according to objective function and some physical and operational constraints in optical networks. The RWA problem is a crucial and complicated designing issue in optical networks which is known to be an NP-hard problem and received much attention by optical network researchers and engineers in literature [8]. The integer linear programming (ILP) [9,10] models have been successfully employed for solving RAW problem for small size optical networks. The ILP based approaches are not feasible for medium and large scale networks due to the increasing complexity of ILP models by increasing network size. Therefore, different heuristic and intelligent algorithms have been developed for solving RWA problem for large scale networks in literature. In [11] the RWA problem was modelled by ILP approach under wavelength continuity constraint and genetic algorithm was employed for obtaining near optimal solutions. In [12] a tabu search based hyper heuristic was applied to RWA problem for minimizing the total bit error rate of the routed lightpaths over WDM optical networks. A systematic form of RWA algorithm using particle swarm optimization (PSO) was proposed in [13]. A RWA algorithm which constructs the routing solution in a distributed manner by means of cooperative ants was reported in [14]. Another ant-based solution for RWA problem was proposed in [15] called ACRWA. The algorithm takes into account both the path length and the congestion in the network to update the values of the pheromone trails. Routing and wavelength assignment in optical networksusing binpacking basedalgorithms waspresentedin[16]. In [17] a delay versus capacity optimization study of RWA problem based on genetic algorithm was presented. The aim of all intelligent approaches is to apply real and practical optimization ideas in nature to RWA problem in DWDM optical networks.