Network design optimization for reliable ethernet ring mesh network

Abstract

Ethernet has been considered to be a predominant solution for recent carrier-class networks due to its simplicity and cost effectiveness. Accordingly, demands for reliable Ethernet networks have considerably increased. A reliable network is capable of not only restoring services in case of network failures as rapid as possible, but also providing a minimum service outage period. In this dissertation, design problems of a reliable Ethernet ring mesh network is investigated complying with the Ethernet ring protection (ERP) introduced by the ITU-T G.8032 Recommendation. In detail, optimization problems regarding network resilience, network availability, and network cost are addressed. As for the optimization of network resilience, we introduce the fundamentals and architecture of the ERP. To evaluate resilience provided by ERP, we implement ITU-T G.8032 Recommendation 2010 at an Ethernet switch and assess network resilience in terms of protection switching time. By the experiments, the problem of long protection switching time is identified in the standard ERP application in heavy traffic congestion, and an algorithm to provide fast and stable protection is suggested as an improvement of the ERP standard. We also implement the algorithm in the switch and experiment performance of the algorithm. We demonstrate that the algorithm provides fast and stable performance even in heavy traffic congestion. Based on the resilient Ethernet ring network, a reliable network design problem in terms of network availability is then investigated. Definitions of reliability and availability are introduced, and network availa-bility model for an Ethernet ring mesh network is provided. We show that the design problem is very complex due to unique characteristics of the ERP. To solve the design problem, two design algorithms, enumeration and heuristic, are proposed with design principles, and it is shown that the enumeration achieves an optimum network design in small or medium scale...