Simulation-based evaluation of a fault-tolerant IP communication system

Abstract:
Automation finds an ever increasing scope of application. Consequently, automation systems and the communication systems deployed within them have to meet new requirements. In safety critical environments, reliable communication is essential in order to prevent situations with severe consequences, like life-threatening scenarios or costly damages. To increase the reliability of communication systems, the concept of fault tolerance can be applied, which enables the system to continue its operation even in the presence of a fault. This is basically achieved by hardware redundancy like redundant links in the context of data communication. Networks with redundant paths entail the necessity of a network protection and recovery protocol. Due to the increasing importance of the Internet, network technologies such as Ethernet and IP find their way even into the automation domain. However, different restrictions and additional demands, such as limited network bandwidth and special real-time requirements, have to be considered in this context. Although many different fault-tolerant communication technologies are available, they are mainly intended for use in high-performance communication systems with a limited number of network nodes. The present work provides an overview of the state of the art of such protocols. Subsequently, an important representative, which is based on a ring topology, is investigated. Since such a system is too complex for a formal analysis and the scalability cannot be examined in a real scenario, a simulation-based approach is used. The communication system under investigation operates at Data Link Layer and is based on Ethernet, the data exchange is based on IP. It offers further Quality of Service mechanisms to categorize the data traffic. The simulation studies are based on networks which consist of links with a bandwidth of less than 1Mbps, operating in half duplex mode. The Media Redundancy Protocol (MRP), a protocol established in the field of industrial automation, is selected and examined under the restrictive conditions stated above. The simulation studies are performed with the discrete-event simulation environment OMNeT++ in combination with the INET framework, which is extended by an MRP model especially programmed for the purpose of this study. The investigations explore the scalability of MRP with various parameter values and the spatial and temporal distribution of the network utilization. Additionally, experiments show how different implementations of ARP influence the network performance. Beyond the results of the simulation, the analysis of the MRP discovers possible improvements for low-speed Ethernet connections.