| Abstract
|
<p>More and more real-time systems a … <p>More and more real-time systems are complex and distributed systems consisting of many sub-systems that must cooperate. In other words, those sub-systems must communicate with each other and to do this, still fullling the overall system specfication, real-time communication must be supported. Moreover, real-time communication is also becoming widely needed in networks like the Internet, spanning over large distances. In for example Internet, the support for different traffic classes is often described as supporting different QoS (Quality of Service) levels and does normally include some kind of specifcation to meet real-time demands. Real-time communication often relies on some kind of scheduling like EDF (Earliest Deadline First), but there are some important differences compared to standard single-processor scheduling. First, the transmission of a packet is, in most cases, non-preemptive. The ongoing transmission, and possibly also some of the already queued messages, can then not be interrupted when a more important message (e.g., with a shorter relative deadline) arrives or is generated. Moreover, a network, instead of a single link, makes the situation much more complex and one must consider things like medium access method, topology, multiple users on multiple nodes, non-deterministic access delay etc.</p>-deterministic access delay etc.</p>
|
