High Performance Fault Management Schemes for Photonic Networks

• Main Authors: I-Feng Huang, 黃一峰
• Other Authors: I-Shyan Hwang
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/16844265990243567167

博士 === 元智大學 === 資訊工程學系 === 93 === ABSTRACT This dissertation proposes a tabular method and a mesh-ring search method for designing fault management mechanisms in Optical Multistage Interconnection Networks (OMINs) and Wavelength Division Multiplexing (WDM) networks, respectively. Limitations of previous literature are discussed, in which heuristic methods are adopted to diagnose crosstalk fault for OMINs. A deterministic algorithm called tabular method is then proposed to record all possible conflicting connections into a table; the connections are then assembled into several conflict-free test set. The first application is used to identify the optimal testing procedure from all possible solutions in the crosstalk fault diagnosis. Results of the proposed method always closely correspond to the optimal fault clusters or minimal test sets for crosstalk fault detection and location procedures. The second application of the tabular method is used to design a fault tolerant mechanism, which is called a fault-free connections scheduling algorithm in the Dilated Benes Network (DBN). The proposed algorithm establishes connections between arbitrary input/output pairs in a minimum number of rounds and averts faulty switches to increase system reliability in DBN. The OMINs can be utilized as major components in the WDM networks, including Optical Add-Drop Multiplexer (OADM), Optical Cross-Connect (OXC), and star coupler, to exchange optical signals. Once the OMINs components in the WDM networks have deteriorated, the entire system performance declines. This dissertation also designs a fault tolerant mechanism for WDM networks. The fault tolerant (commonly referred to as fault recovery) mechanism for WDM networks can be categorized as either protection or restoration. In the protection mechanism, each connection reserves backup paths statically during call setup. In the restoration mechanism, each connection that traverses a failed block discovers dynamically an adaptive backup route after failures occur. For various fault protection requests, the protection mechanism can be either a dedicated or shared facility. Depending on where a detour originates, the fault recovery mechanism can be classified into either link, path or segment recovery method. While previous studies addressing fault recovery issue in WDM networks largely focused on the link and path recovery, segment recovery in WDM networks has seldom been investigated. Networks have become increasingly complex given the accelerated demand for a larger bandwidth. Mesh networks are widely considered to be flexible, scalable, bandwidth-efficient, cost effective, and dynamic. Therefore, this dissertation investigates segment recovery schemes in the WDM mesh networks. To consider their feasibility, a mesh-ring search method is proposed to derive the adaptive recovery paths. Each node in the networks searches independently for its own mesh-rings; these mesh-rings then become adaptive recovery paths after calculation. The first application is used to design a segment protection mechanism in WDM networks, and then two schemes are presented. The first scheme in this application is called Dynamic Shared Segment Protection (DSSP). This algorithm uses pre-calculated mesh-rings to divide the working path into several working segments; resources are then reserved to protect these working segments. The proposed DSSP algorithm can utilize resources efficiently and increase the rate of successful protection, making it applicable in arbitrary mesh networks. The second scheme in this application is in the implementation perspective of segment protection, which is called Overlap Segment Protection (OSP) and Non-Overlap Segment Protection (NOSP). This scheme also applies pre-calculated mesh-rings to construct adaptive protection segments, which can or cannot overlap each other. The second application is used to design a fault restoration mechanism called Dynamic Multiple Ring Algorithm (DMRA) for the WDM mesh networks. DMRA pre-computes the mesh-ring architecture before faults occur enabling the proposed DMRA algorithm to recover quickly from failures and determine how to allocate recovered traffic loads based on the current traffic load and the network bandwidth along the restoration paths. The third application integrates previous applications to design a recovery mechanism called Guaranteed Quality of Recovery (GQoR). Four classes of priority for GQoR mechanism are applied based on the customer’s request, with each one mapped to the adaptive recovery method. The GQoR mechanism confirms that customers possess an efficient recovery time and sufficient backup resources. The above fault management methods in photonic networks are introduced, developed, and simulated herein. Simulation results demonstrate the feasibility of applying the proposed methods in real-world networks.