Date of this Version
E-science projects require very high-speed and reliable networks to transfer data across various destinations in the world. Dynamic Circuit Network (DCN) is a networking service to make advance reservation of bandwidth between a source and a destination in a network. In this thesis we solve the problem of advance reservation of bandwidth in next-generation wavelength-division multiplexing (WDM) networks using a simulation based approach.
We implement a greedy algorithm and a genetic algorithm in parallel, in separate threads. The request for advance reservation is processed by both but the user gets the response only from the greedy algorithm. The genetic algorithm is used for offline re-optimization where we optimize the schedule of all the future reservations thereby maximizing the number of reservations possible in the network. We evaluate the approach using trace-driven traffic and simulated traffic. We observed an improvement in the blocking probability and service blocking probability of the network using our approach.