Recently, Internet traffic is rapidly growing due to the emergence of new applications (IPTV, video on demand). According to the increase of IP traffic, network equipment with large capacity should be deployed in the optical networks. However, the upgrade of network equipment leads to increasing the energy consumption of the network.
The optical networks based on Wavelength division multiplexing (WDM) technology are widely deployed in backbone networks because of their large capacity over long distance. In order to accommodate the increase of traffic, a link can be composed of multiple fibers in WDM network. On the other hand, network with Mixed Line Rate (MLR) (10/40/100 Gbps) is one of promising candidates to support heterogeneous traffic demands.
The routing and wavelength assignment (RWA) is the problem of setting up lightpaths by routing and assigning a wavelength to each connection. Traditionally, the goal of the RWA problem is to minimize the load on available resources, in order to minimize the blocking probability of new requests. However, this leads to a waste in the power required to keep up and running both OXCs and optical amplifiers along fiber links.
In this paper, we propose the dynamic RWA algorithms to improve energy efficiency in IP over WDM networks. We consider two network architectures, multiple fiber links and MLR. To improve the energy efficiency in the networks with multiple fiber links, the proposed algorithms use the V-like cost function to decide the fiber cost according to the number of used wavelengths. And the layered graph is used to select the energy efficient route and wavelength. The proposed algorithms in MLR networks use the cost function considering not only the energy consumption, but also the Quality of Transmission (QoT) caused by transmission range constraint.
The proposed algorithms are compared and analyzed with conventional algorithms in view of average power consumption and blocking probability through OPNET modeler. Results indicate that proposed algorithms can save the power consumption compared with conventional algorithms, without the increase of blocking probability.