Suethanuwong, E. (2011). Performance evaluation and optimization of standard-ethernet traffic in TTEthernet systems [Dissertation, Technische Universität Wien]. reposiTUm. https://resolver.obvsg.at/urn:nbn:at:at-ubtuw:1-60071
Nowadays standard Ethernet is widely used in home and office computer networks, and well recognized as the low-layer protocol of the Internet protocol. There are three main advantages of using today's Ethernet technology: (1) high bandwidths, for example with Fast Ethernet and Gigabit Ethernet, (2) open standard protocol, defined in IEEE 802.3, and (3) low prices of commercial-off-the-shelf standard-Ethernet devices. In additional, today's Ethernet technology, which is so-called switched Ethernet, has higher bandwidth utilization than traditional Ethernet based on bus network topology, since switched Ethernet provides full-duplex communication and no collision occurrence. However standard-Ethernet, which was not originally designed for real-time communication, does not provide a timely-determinism property. In the last decade there have been many designs that adapt standard Ethernet to provide temporal guarantees. Such a standard-Ethernet based protocol augmenting with temporal guarantees is called real-time Ethernet (e.g.<br />Ethernet POWERLINK, SERCOS III, PROFINET IRT, Ethernet/IP, MODBUS-TCP, AFDX, and TTEthernet).<br /> Recently there has been a main driving force behind applications of real-time Ethernet that is the increasing demand for a seamless connectivity between legacy Ethernet networks and real-time Ethernet networks. This can be seen in the industrial automation domain, for example, the connectivity between the administrative tasks of a company (office networks) and the factory floor (fieldbus networks) has been increased. Another main driving force for adapting standard Ethernet to real-time Ethernet is to integrate all application types into a single standard-Ethernet based network. Such a single standard-Ethernet based network is potentially very cost effective, and reduces the complexity of network infrastructure.<br />TTEthernet was designed to be a novel unified communication architecture that meets the requirements of all application types from non real-time applications, to real-time applications, to the most demanding safety-critical real-time applications. TTEthernet were aimed to altogether achieve the three main properties: (1) determinism property for real-time traffic, (2) full standard-Ethernet compatibility for real-time communication, and (3) high Ethernet-bandwidth utilization for standard-Ethernet traffic. Due to the fact that TT traffic in TTEthernet systems takes precedence over standard-Ethernet traffic, the flow of standard-Ethernet traffic depends only on a TT-traffic communication schedule. The objective of this thesis is to investigate a TT-traffic scheduling approach for obtaining the highest performance of standard-Ethernet traffic coexisting with the TT traffic in the same TTEthernet systems. The throughput and timing-performance of standard-Ethernet traffic are evaluated in both simulated and physical 100-Mbps TTEthernet systems. For the simulated TTEthernet systems a simulation model for TTEthernet systems is developed and implemented.