Research Of PCE-Based Dynamic Routing And Resource Computation Approach In Multi-Region Intelligent Optical Networks

Transport network is critical in the unsplit telecommunications networks, which provides transparent circuit-switched paths for long-haul transport service from end to end, and carries the flow generated from the IP/MPLS router in the core data communication networks. Based on the backbone transport networks, the intelligent optical network, represented by the Automatically Switched Optical Network (ASON), is a smart management and control system platform attached to the optical switching elements and transmission devices. It is the natural trend of the development of transport-network technologies and the interconnection of heterogeneous backbone networks, to provide the emerging dynamic, flexible and fine-granularity path service directly on the optical layer.Since routing and resource assignment strategy may affect the actual capacity of circuit-path services, it is worth designing effectual routing and resource computing algorithms to improve the network-traffic performance, but nevertheless dynamically switched optical networks may not afford too much computing time especially in multi-constraint scenarios. As a result, PCE architecture has been proposed to centralize the path computing function and to fully utilize the hardware advantages of high-performance servers and power workstations in the cloud, which has already been a studying hot point for both the academic and engineer circles. In the year 2010, the Dual Routing Engine (DRE) architecture was proposed to provide an enhanced control plane for the intelligent optical networks by combining the GMPLS and PCE architecture primly. To be aimed at the enhancement of dynamic multi-region routing technologies in the DRE architecture, this academic dissertation elaborates the DRE platform environment, PCE-based multi-region routing designing and implementation scheme, and the Message Policy Analyzer (MPA) agent module for multi-region path computation.The main contributions of this dissertation are listed as follows: Firstly, a distributed software prototype of large-scale extensible multi-layer and multi-region control plane DREAM is developed, which in part supports GMPLS/PCE/PCEP protocol suites. This dissertation specifies the functional designing and architecture of Group Engine (GE) in DREAM platform, as well as the object-oriented designing of event-driven MPA module by taking reactor-based synchronous I/O multiplexing mechanism.Secondly, multi-region path computation schemes including BRPC, DRPC, as well as the two derivative schemes of Per-Domain are expounded in the dissertation, with the complete implementation solutions of message processing procedure inside the MPA module of DREAM platform. Furthermore, the dynamic determination of "Stitching PCE" is proposed in the DRPC scheme.Thirdly, supported by the DREAM platform, a new integrated hierarchical PCE (IH-PCE) model has been proposed, under which a new procedure named "Source-driven Hierarchical PCE-based Computation (SHPC)" is worked out to compute the global shortest paths in the multi-region environment. Experimental result shows that SHPC achieves a lower blocking probability in the multi-region optical networks under the scheme of "Routing with Distributed Wavelength Assignment (R+DWA)" than Per-Domain and the non-hierarchical PCE-based schemes such as BRPC and DRPC. Besides, SPHC applies to the scenario of domain-sequence loop, under which cross-layer cross-region load balance is possible, while the other schemes fail in the domain-loop scenario.