Research On Routing, Resource Allocation And Scheduling In Intelligent Optical Networks

The rapid development of various applications including IP broadband services, mobile Internet and the network of things applications leads to the explosion of bandwidth demands and necessitates the ultra-large capacity, dy-namic and flexible next generation optical transport networks. Various applica-tions and services also demand efficient and flexible network resource control and management.Current wavelength routed optical networks have several drawbacks in supporting such applications including coarse granularity, fixed rigid grids, poor scalability of control and management and relatively high operating ex-penditure. With the introduction of bandwidth-variable data transmitting and flexible optical switching technologies, elastic optical networks can tailor the allocated spectrum width to the specific request needs hence improve the re-source utilization. Apart from the demand of advances in the underlying op-tical transmission technologies, the next generation optical transport networks require a more flexible, intelligent and scalable network control and manage-ment solution. GMPLS-based unified control plane, which is composed of a set of protocols for routing, signaling, etc., ensures intelligent routing and resource allocation in optical transport networks. Path Computation Element architec-ture provides path computation for flexible, multi-constraints, complex routing requests. Appropriate utilization of them can provide flexible dynamic traffic provisioning and scheduling while accommodate as many traffic requests as possible.To cope with the challenges of the next generation intelligent optical net-works, this thesis mainly investigates flexible resource allocation and efficient network management. The research of this thesis, including routing, mod-ulation format and spectrum allocation in elastic optical networks, conges-tion avoid routing in wavelength routed optical networks, CapEx optimized scheduling in multi-layer optical transport networks and network scalability e-valuation in multi-domain networks, etc. is supported by National863projects. The main innovative contributions of this thesis are listed as follows.Firstly, to solve the complex routing and spectrum allocation (RSA) prob-lem in elastic optical networks, especially the modulation format allocation problem, the author proposes three spectrum efficient routing, modulation for-mat and spectrum allocation algorithms (Link-MLC, etc). Simulation results show that the Link-MLC algorithm reduces blocking probability and increases resource utilization rate.Secondly, to resolve the load balancing problem in large optical transport networks, an extended Path Computation Element (PCE) architecture is pro-posed along with two congestion evaluation methods. Simulation results show that both algorithms can reduce the blocking probability and achieve load bal-ancing.Thirdly, the author looks into the network scalability evaluation problem, especially the evaluation of OSPF-TE convergence time in large multi-domain intelligent optical networks and points out the main contributor of the conver-gence time is the information flooding procedure. The devised convergence time measuring methods are tested for validation and implemented on multi-layer multi-domain optical network testbed.Fourthly, to cope with the soaring capital and operation expenditure for next generation optical transport networks, the author proposes two optimiza-tion methods (Transponder/Regenerator Reuse (TRR) approach, etc). Simu-lation results show that TRR can reduce network cost. And the tabu search heuristic performs better than the greedy heuristic when the same approach is used.