Abstract
The rationale behind a green network is that it should effectively reduce energy consumption, while maintaining the level of services for data communications. In this paper, we propose an efficient approach, called the Compression Algorithm (CA), which is designed to solve the link on/off and weight assignment problems jointly so as to minimize a network's energy consumption. The problem is formulated as a mixed integer non-linear optimization problem. Because the problem is NP-hard, the CA utilizes a genetic algorithm to determine the link on/off schedule. In addition, it exploits the simulated annealing technique for link weight assignment so that the routing paths satisfy the link capacity constraints. By solving the link on/off and weight assignment problems sequentially, the CA scheme reduces the uncertainty about network energy consumption and yields a near optimal solution. To observe the relationship between network energy consumption and link load distributions, performance evaluations were conducted on three schemes, namely, the proposed CA, route construction without considering power savings, and route construction using minimum power saving without link capacity constraints. Numerical results demonstrate that the CA outperforms the other approaches on a network embedded with both uniform and non-uniform demand distributions. © 2012 Elsevier B.V. All rights reserved.