Research On Survivability Of Multi-granularity Traffic Grooming Optical Network

Intelligent optical network (ION) imports GMPLS based control plane which supports multi-granularity switching and dynamic bandwidth resourse provisioning capability. It introduced intelligence and flexibility into optical networks. Survivability of multi-granularity optical network (MG-ON) has been widely agreed as one improtant technology in ION. This thesis studied survivability of MG-ON with traffic grooming ability via theoretical analysis, systematic modeling and performance simulations.Firstly, this thesis gives a brief introduction of the development and applications status of optical network, along with the background analysis of survivability issue in MG-ON. Then, some fundamental pricinples are presented including node architecture, GMPLS control plane, and multi-granularity cross-connect, etc. Major survivability strategy of GMPLS failure recovery process, traffic grooming, and survivable traffic grooming are given.Secondly, this thesis focuses on single-link restoration mechanism in the GMPLS-based MG-ON. Based on the analysis of traditional multi-granular restoration mechanism including waveband path restoration for waveband failure (WBRWBF)and wavelength path restoration for wavelength failure (WRWF), an improved GMPLS-based shared restoration (GSR) mechanism for MG-ON is proposed. Once link fails, GSR adopts differentiate recovery strategy for different granularity failed types. Theoritical and simulation shows compared with conventional WRWF + WBRWBF recovery method,GSR have significantly performance increasement for successful restoration rate. Moreover, the greater the granularity of waveband G is, the fewer node ports network needs.Finally, multiple concurrent failures protection algorithm with traffic grooming is explored. For multiple concurrent failure of shared path protection optical network, an improved resourse reprovisioning algorithm—alternative working path reprovision first algorithm (AWPRF) is proposed. AWPRF algorithm can reduce the total number of unprotected and vunerable requests whihc need to be reprovisioned. Simulation is executed for comparison between AWPRF with traditional GBR and GBR-ILP algorithm. Simulation results show that AWPRF algorithm caused less unprotected and vulnerable request than conventional GBR algorithm. At the same time, AWPRF algorithm achieved restoration performance compared to GBR-ILP algorithm while network load is light. When network load is heavy, AWPRF algorithm has a little worse restoration performance with obvious algorithm complexity reduction.