Research On Optical Multicast Packet Switching Node Architecture And Scheduling Algorithm Based On FDL

Emerging services, such as High-Definition TV (HDTV) or Video on Demand(VOD), make that the amount of network multicast traffic increases rapidly. Whenmulticast services are transferred in switching nodes, optical packets need to be copiedand forwarded. Whereas, copying and forwarding multicast packets in optical switchingnodes often lend to the aggravation of wavelength contention. Therefore, in order toresolve the wavelength contention, it is necessary to design appropriate node architecturesand appropriate multicast scheduling algorithms for optical multicast.Fiber delay line is the device that is nearest to the electric random access memory inresolving contention of optical packets. In chapter2, the buffering mechanism of opticalpackets in FDLs is studied, in detail, the classification of FDL buffering, the queue systemmodels and the scheduling policies of optical packets in FDLs are analyzed.In chapter3, a novel node architecture that is used to resolve contention of opticalmulticast packets is designed. The node architecture is composed of N input ports/outputsports, optimal switching module, controlling module, network coding module and fiberdelay line loop shared feedback buffering module. Since the all-optical logic XOR gatebased on integrated SOA-based Mach-Zehnder interferometer is proved to be used toresolve wavelength contention, in the light of the all-optical logic XOR gate, networkcoding module is designed for contention resolution in chapter3. Meanwhile, the fiberdelay line loop shared feedback buffering module is configured in the node architecture toresolve contention in time domain. As for the optical multicast scheduling in the fiberdelay line loop shared feedback buffering module, a minimum length cascaded bufferscheduling technique is designed, and a whole scheduling algorithm based on thescheduling technique is designed. The simulation results show that the minimum lengthcascaded buffer scheduling technique can effectively decrease the buffering delay whencompared to the existing algorithms, and the designed node architecture is better indecreasing packet loss probability and reducing the average delay when compared to theexisting node architecture we have introduced.For the packets scheduling problem in the node architecture which take the multicastenabled fiber delay lines as input buffering, the existing scheduling algorithms cannotachieve the optimal scheduling strategy, intelligent optimization algorithms have not beenused for packets scheduling in the node architecture. In chapter4, the genetic algorithm is introduced in packets scheduling problem in the node architecture. Chromosome structuresuit for the actual problem and the corresponding constraint conditions are designed.Furthermore, an improved genetic algorithm is proposed combining the thought of theimmune genetic algorithm and the genetic simulated annealing algorithm. The simulationresults show that when compared to scheduling algorithms based on the basic geneticalgorithm and the immune genetic algorithm, on the premise of loss less packets, thescheduling algorithm based on the improved genetic algorithm can get smaller totalbuffering delay, that is to say the scheduling algorithm based on the improved geneticalgorithm has better performance.