Protection And Traffic Grooming In Optical Networks

Optical networks based on wavelength-division multiplexing (WDM) technology offer the promise to satisfy the bandwidth requirements of the Internet infrastructure, and provide a scalable solution to support the bandwidth needs of future applications in the local and wide areas. Using WDM technology, an optical fiber link can support multiple non-overlapping wavelength channels, each of which can be operated at the speed of 10 Gbps or40 Gbps today. Therefore, the resource allocation problem including routing and wavelength assignment is becoming more and more important. It will endeavor to satisfy the demands and their QoS requirement. In Chapter 2 and 3, this paper introduces the resource allocation problem in a single network. In Chapter 4, it introduces the problem in a multinetwork.Since the main advantage of optical network is its large bandwidth the destruction of fiber and/or other equipments in nodes will undoubtedly lead to mass demands losing. The protecting strategy is used to keep the communication fluently. In Chapter 2, the paper constructs the ILP protection model for routing and wavelength assignment. For large-scale network, Bender decomposition is proposed. The result from computer shows that ILP model can provide the optimal solution. But the problem belongs to a NP-C problem, which implies that the time to solve the problem for large-scale network is too long and the memory of computer would run out to get the feasible solution. Decomposition theory proves to build an analytic foundation for design of modularized and distributed control of networks. In theory the decomposition gives the same solution as ILP model does. However, the test validates that the solution of decomposition is near-optimal and requires less time and smaller memory than ILP model.In time-slot WDM network, the communication is carried by frame. And each frame can be divided into slots. Contrast to the un-slot WDM network using the wavelength-granularity, the slot WDM network could build more connection under the same condition. In Chapter 3, with the static demands the paper proposes the ILP slot modle. The test shows that ILP model could get the optimal solution. But it could fail in large-scale networks without any decomposition. In the WDM network, only a fraction of customers are expected to have a need for such a high bandwidth of wavelength. So it is not economic to provide a single wavelength for them because of utilization. In order to save network cost and to improve network performance, it is very important for the network operator to be able to groom multiple low-granularity demands onto high-granularity circuit pipes. In practical network traffic, there are no only many demands for node-to-node connections but also various physical constraints. In Chapter 4, the paper proposes packing grooming algorithms based on partial combination and max revenue. The results show that algorithms can be very effective to deal with physical constraints within demands in traffic-grooming problem. And they have less resource consumptions and more practically in large- scale networks.Chapter 5 concludes the whole paper and future work.