A genetic algorithm for solving virtual source placement problem on WDM networks

As WDM technology matures and multicast applications become increasingly popular, supporting multicast at the WDM layer becomes an important and yet challenging topic. In order to reduce the number of wavelength channels to achieve the multicast effectively, switching nodes with light-splitting and wavelength-converting capabilities denoted as virtual source (VS) nodes are developed. In this paper, given a WDM network, a positive integer k and a set of multicast requests, the VS placement (VSP) problem on WDM networks is studied; the goal is to determine the locations of the VS nodes, the multicast routing and assigned wavelengths of multicast requests so as to minimize the number of used wavelength channels. Since the VSP problem is a hard problem, a genetic algorithm (GA) is proposed to solve it. In the proposed GA, a binary-bit array is used to represent the locations of the VS nodes on network. For a given locations of VS nodes, three multicast routing methods: core-based tree (CBT), link-disjoint CBT (LDCBT), and layered graph (LG) are proposed and used to construct the shared tree for multicast requests. In the CBT and LDCBT methods, a multicast tree constructing (MTC) algorithm is used to construct the multicast tree of a given multicast, and a segment-based wavelength assignment (SBWA) algorithm is proposed and used to determine the assigned wavelength of the multicast tree. Moreover, in the proposed GA, several crossover and mutation operators are developed and used to generate offspring. Simulation results show that the proposed GA together with LG or CBT multicast method can get better results.