Research Of Pce And Routing In Multi-Layer Multi-Domain Intelligent Optical Networks

Traffic engineering (TE) effectively improves resource utilization in GMPLS/ASON multi-layer multi-domain optical networks. In single domain networks, traffic engineering information such as link state and wavelength availability is broadcasted by internal gateway protocol (IGP), e.g. OSPF-TE, and label switched path (LSP) can be computed by the local source node or remote PCE. However, in multi-domain networks TE information such as wavelength is limited to intra-domain because of considerations of network extensibility and scalability. Inter-domain LSP cannot be effectively computed because the limit of traffic engineering information around the area boundaries. The PCE based architecture proposed by IETF solved this problem by deploying one or more PCEs per domain. But the domain sequence is known as a prior to compute the optimum LSP.In hierarchical PCE (H-PCE) based architecture, hierarchical relationship such as parent and child is introduced between PCEs. The parent PCE is responsible for maintaining the inter-domain connectivity and traffic engineering information, while each local child PCE is responsible for their own intra-domain LSP computation. This architecture allows optimum domain sequence to be selected. Furthermore the parent PCE can ask child PCEs to compute concurrently respective LSP segments across their domain to improve the computation efficiency.This dissertation researched into the application of PCE in the routing of multi-domain and multi-layer networks. The main contribution including innovative research, design and development work is listed as follows.Firstly, in-depth research has been conducted on the PCE related routing technologies and a complete, standard and flexible PCEP protocol library was developed. On this basis, PCC and PCE as well as some common path computation algorithms were developed. Then the PCE was embedded into the original simulation test platform, completing the evolution to the PCE-based architecture.Secondly, the PCE location programming problem in H-PCE based architecture was proposed based on the analysis of path establishing process in hierarchical PCE based networks. A mathematical model of comprehensive communication cost was established and specified.Thirdly, a quantum discrete particle swarm optimization (PSO) based PCE placing method was proposed to minimize the comprehensive communication cost to the global approximate optimal solution. The simulation results show the improvement of path establishing delay and blocking probability.Fourthly, a software control plane simulation platform according to GMPLS/ASON related standards was designed and developed. The author led to design and forged the fundamental network environment based on virtualization technology and responsible mainly for the connection controller (CC) module development.