Study On Technologies Of Survivability And Intelligent Control Plane For Optical Networks

With Generalized Multi-Protocol Label Switching (GMPLS) based distributed control plane, Intelligent Optical Network (ION) can provide dynamic network resource discovery capability, automatic connection establishment / teardown capability, traffic engineering, and efficient protection / restoration capabilities,etc. This dissertation presents a systematical and in-depth study on survivability issues in ION, GMPLS/OBS hybrid network control plane technology, survivability of Passive Optical Network (PON), and dynamic bandwidth allocation technologies via theoretical analysis, systematic modeling, algorithm construction, and performance simulations.Firstly, this dissertation studied survivability in Multi-layer and Multi-domainoptical networks with the brief introduction of survivability issues in optical networks. An improved multiple granulariy traffic grooming supported mechanism called GMPLS Shared Recovery (GSR) algorithm is proposed. Simulation shows it has better successful recovery ratio. An Alternative Working Path Reprovision First (AWPRF) algorithm is proposed for backup resource reprovisioning which has better recovery performance than the traditional GBR and GBR-ILP algorithms. An improved ant colony based DBA algorithm called SA-DRWA is proposed to solve the Routing and Wavelength Assignment (RWA) problem. Simulation shows both the blocking probability and resource utilization is improved with the introduction of random perturbations and link idle ratio as constraints.Secondly, this dissertation presents study on hybrid GMPLS / OBS network control plane technologies. Based on the overview of general survivability model for control plane, the control plane recovery time model for performance evaluating is provided with performance analysis of typical protection strategies. An improved signaling mechanism for hybrid GMPLS/OBS network is proposed. Simulation shows the improved mechanism reduces link establishment delay compared with standard RSVP-TE protocols. A QoS based LSP Shared (QLS) mechanism is presented for improving the QoS performance of hybrid networks. Theoretical and simulation analysis shows the QLS mechanism obtains lower packet loss of low level service while the performances of high level service are guaranteed simultaneously.Finnaly, this dissertation presents study on survivability and dynamic bandwidth allocation in Passive Optical Network (PON). The recovery model of two-stage cascaded PON is presented and simulated to evaluate performance with different split ratios and protection strategies. An improved upstream transmission scheme with non idle loss and the dynamic bandwidth allocation algorithm called NIL-DBA are proposed. Theoretical analysis and numerical simulation show that NIL-DBA algorithm can effectively improve resource utilization and reduce service delay.