A novel dynamic fault restoration mechanism using multiple rings approach in WDM mesh network

Abstract

This work proposes a distributed algorithm for fault restoration, called the dynamic multiple rings algorithm (DMRA), in a DWDM mesh network. This study explores how to choose restoration paths and distribute a fault-tolerant bandwidth when link, node and channel faults occur. Mesh network topology is becoming increasingly popular, and includes many restoration paths. Calculating the appropriate restoration paths will take a long time if the whole network is considered. Accordingly, the basic idea of this paper is trying to use networking segments near faults to share the restoration load in mesh networks. Each node is responsible for searching restoration paths around it using the proposed DMRA. Nodes use the distributed algorithm to search nearby nodes and establish the relationship between them to build many logical rings. Nodes also can locate the fault in which logical rings are, and these logical rings will establish restoration paths. The traffic load over faulty links or nodes can be diverted to other paths in the networking segments. The cost of restoration paths is computed at each node from both current capacity and transmission time. The selected restoration paths are the appropriate transmission routes in the nearby networks. Accordingly, searching restoration paths and assigning wavelength routings can be performed quickly, depending on network bandwidth and the traffic load. Furthermore, the proposed algorithm is adapted to dynamic networks, and the control messages are to be transmitted in bi-directional control channels. The proposed fault-tolerant mechanism is designed not only for a single fault, but also for multiple faults in the network. Improving the stability and load balance of the network is therefore helpful. Therefore, the method proposed herein can be adopted in a large-scale and complex network topology. The simulation environment employed in this paper is based on the architecture of NFSNet, which includes 14 nodes and 21 links. There are 10 wavelengths in each link. Each node can compute the multiple rings architecture, and some nodes will start up to broadcast to search for a multiple rings architecture in a period for the entire network topology. The traffic loads are simulated dynamically and set from 10% to 90% of entire network utilization. The blocking probability of restoration paths using DMRA is compared with that using p-cycle in various traffic loads when a fault occurs. The results show that the blocking probability using DMRA is lower than that using p-cycle. So, it is obvious that the performance of DMRA is better than that of p-cycle. Other relevant issues, such as fault restoration in GMPLS, burst switching, and load balance, are under investigation.