Study of Dynamic Topology Change for Total Energy Consumption in Green IP Networks

In this work, we investigate the routing and dynamic link on/off management for green IP networks. Two approaches are evaluated: Global Topology Change (GTC) and Local Traffic Redirect (LTR). In GTC, the on/off state of each link is known by all of the routers in the network. Any link state change will trigger a network topology update in each router. In the transient period link state change, the inconsistency of topology images among different routers would result in traffic loops. In LTR, the link on/off states are only known by the owner of those links. The image of global network topology is remained fixed for each router no matter how the link state changes. The benefit of this approach is that the network topology is kept the same in each router. However, each router has to carefully maintain its own routing table to manage the flows going through it. The contribution of this paper includes four parts. First, we evaluate GTC and LTR to determine which is better for achieving green IP networks. Second, we address the dynamic control mechanism of link on/off states so as to alter network topology for power saving. A threshold-based scheme is proposed to determine the timing of changing link states. The third contribution comes from that we propose a protocol for updating link on/off states for the GTC approach. This protocol gives networks immunity from the traffic loop problems. It also prevents the network from being disconnected when multiple routers turn off their links at the same time. Since network survivability is mandatory to guarantee network integrity, in the fourth contribution of this paper, we propose an IP protection scheme to guarantee 100% network survivability against any single node failure while maintaining energy efficiency.