Research On The Routing And Fault In Multistage Multi-plane Optical Switching Fabric

Traffic diversification and explosive growth of traffic capacity have made thetransport network facing with challenges of high quality, high survivability, flexibleconfiguration, intelligence and packetization. The fast improvement of the opticaltransmission network put more requirements on the network nodes. However, in theconventional network nodes, the OEO processing and the electrical switching bringproblems such as high power consumption and difficult temperature control as the nodecapacity increases. Optical switching fabric can reduce the power consumption and easethe temperature control. Moreover, the multistage multi-plane Clos structure-basedoptical switching fabric has other benefits like scalability, modularity and low blockingrate.Though the routing algorithms for electrical switching fabric can also be used withoptical switches, necessary modifications have to be made due to some specialconstraints from the optical devices. Meanwhile, the existing routing algorithms do nottake into account the effects of faults which have some new properties as the size of thefabric expands. Considering the above problems, the contribution of the dissertation isas follows:Firstly, the routing algorithm is applied to the rearrangeable multi-slot multistagemulti-plane Clos optical switching fabric where the PAULL algorithm is used forrearrangement. The resource utilization rate and rearrangement inside the plane andacross the planes are both studied. Then, considering the constraint from the opticaldevices, a modification to the routing algorithm is introduced and its performance isanalyzed.Secondly, the characteristics and effects of faults in the multistage mult-plane Closfabric are studied. Combined with the time-division feature, a fast fault localizationmethod based on matching public resources is proposed. Simulation results show thatthe method can achieve a better result under a certain number of time slots.Finally, according to the different application scenarios, limiting algorithm andmultipath fault protection methods are proposed respectively. Considering the combined fault effects to different layers, the limiting algorithm limits the traffic in the outmostlayer to achieve efficient scheduling of the following layers. Simulation results showthat the proposed limiting algorithm can reduce the blocking probability effectively,increase the computing efficiency. Multipath fault protection method can use eithercomplete rerouting or partly rerouting. Simulation results show that partly rerouting hasa relatively high failure probability due to limited available resources. Instead, thecomplete rerouting can reach a higher success probability at a cost of more complicatedscheduling. A combination of the two methods is a better choice.