The Step-by-Step Virtual Topology Reconfiguration Using MDP in the WDM Network

topology is being reconfigured to be more adaptive. It can reduce network congestion flow, and the ratio of information loss. So, the efficiency of network is improved[1,2]. e study the issues when considering the problem of configuring WDM networks in response to changes the traffic patterns. As traffic patterns change, rtual topology of network is reconfigured to adapt the affic requests. The problem is that network is halted ring the reconfiguring the virtual topologies. erefore the delay of reconfiguration is important. econfiguration can be performed at one step or veral step. Then, there is no research considering the quence of reconfiguring. To maintain the rvivability of networks, and to adapt to the dynamic affic patterns, we have to make decisions about quence of reconfiguring virtual topology. In this per, we formulate the problem as a Markovian ecision Process(MDP). We apply results from MDP eory to obtain optimal reconfiguration policy. But, the batch process of reconfiguration causes disadvantages which reduce efficiency of the network. The major reason is that the network is halted during the virtual topology reconfiguration process. If the buffer of each node is not large enough to store whole packets or it takes long time to reconfigure, most packets will be overflowed[2]. Naturally it will cause the loss of information and the average packet delay will be increased. For those reasons, researchers have studied the step-by-step method of virtual topology reconfiguration which performs transitions by several steps. When it comes to this kind of approach, there are some reconfiguration schemes using branch exchanges[2] and many studies focusing on minimizing the number of retunings at each node[3]. eywords: MDP, Reconfiguration, Survivability, rtual topology, wavelength assignment