A PWCE based Fast Rerouting Approach in MPLS Traffic Engineering for Link and Node Restoration

• Main Authors: Sheng-Yi Chang, 張聖翊
• Other Authors: Huan Chen
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/18066228479218865126

碩士 === 國立中正大學 === 電機工程所 === 93 === Fast rerouting the specific working label switched path (w-LSP) is an essential feature of traffic engineering in MPLS networks to deal with link and node failure restoration problem. This work integrates fast rerouting with a novel concept of Protected Working Capacity Envelop (PWCE) to achieve link and node failure recovery over reserved network in simple and efficient way. Thus, the basic fast rerouting function for w-LSP is often performed at custodial nodes between the affected link or node failure along the primary path. The restoration path is formed within the alternated backup LSPs (b-LSP) whose links represent the spare channels with bounded hop count limit. Our aim is to resolve this local restoration problem for link and node failure recovery by fast rerouting demands to a pre-planned restoration path. In this paper, the PWCE concept is introduced with traffic engineering to enhance fast rerouting mechanism in survivable MPLS networks. We consider two dominant scenarios of failure in MPLS networks: single link failure and single node failure. PWCE is characterized by a variable index (volumes) as spare capacity configured on physical link for fast rerouting w-LSPs to alternate b-LSPs. We first formulate this survivable routing and spare capacity allocation problem as a mixed integer non-linear optimization programming (MINLP). Because the MINLP problem is NP-complete in those scenarios, we then develop an approximation algorithm by Lagrangean relaxation with Heuristics (LRH), aimed to resolve arc weight assignment and capacity configuration to protect single link and/or single node failure in MPLS networks. The task is first formulated as a mixed integer non-linear optimization problem in which the bottleneck link blocking probability is to be minimized. Numerical results demonstrate that LRH outperforms minimum working hop approach in blocking probability for networks with uniform and non-uniform demand distributions.