Research On Traffic Grooming In Optical Networks

With the advent of the era of multimedia and continuous growth of network traffic,the emergence of new types of applications greatly enhances the bandwidth requirements.It’s imperative to increase the capacity of the existing networks to sustain such a growth.We see the trend of keeping on deploying new equipments in the backbone networksaround the world, but the high price of network equipments slows down the pace of suchan expansion. How to reduce the network cost has become an important research topic.Wavelength Division Multiplexing (WDM) technology can multiplex and transmitsignals at different wavelengths in the same fiber, and the characteristics of largetransmission capacity, technical adaptability, and ease of implementation enable it to beone of major transmission technologies in today’s backbone networks. However, thefrequency of each WDM signal carrier is fixed and frequency spacing between twoneighboring channels is coarse, which lead to a low fiber spectrum utilization. In contrast,through using multiple subcarriers, the orthogonal frequency division multiplexing (OFDM)technology splits a high-speed data stream into a bunch of low-speed data streams, andthrough working with the flexible grid technology, can greatly improve the fiber spectrumutilization. Thus, the OFDM-based optical network is considered as a promising candidatefor the next generation optical networks.Based on WDM and OFDM optical transmission techniques, this thesis studies thetraffic grooming problem for an optical network to reduce the total number of requiredoptical transponders, and in turn the network cost. The key research contents in this thesisare summarized as follows: First, for the WDM optical network, we develop a mixedinteger linear programming (MILP) model to minimize the total number of requiredtransponders. Second, due to the computational difficulty in solving an MILP model, we also present an efficient heuristic algorithm for the traffic grooming problem to obtain asub-optimal solution for a large network. Third, we develop an MILP model for the trafficgrooming in the OFDM-based optical network. Finally, we solved the MILP models andrun simulations for the heuristic algorithm. We also compared the performance in terms ofthe number of required transponders between the cases of without traffic grooming andwith traffic grooming. We verified the advantage of traffic grooming in terms of thenumber of required transponders in the entire network.