Browse Theses

The real-time communications problem is distinguished from the standard nonreal-time problem by the introduction of time. In the nonreal-time setting, it is sufficient to verify the functional correctness of a communications solution along with some notion of aggregate throughput and/or average delay. In the real-time setting, the individual message-level timing requirements are an integral part of the functional correctness. The current state of the art and predominant practice of using Time Domain Multiplexing (TDM) techniques to insure timing correctness tends to result in ad hoc, handcrafted timing solutions which are very hard to modify, upgrade and extend. Further, the deadlines of synchronous communications are guaranteed by relegating asynchronous class messages to a noninterfering, background status. This approach often results in poor system responsiveness. This thesis develops a novel approach for scheduling the IEEE 802.5 and SAE-9B token ring LANs for hard real-time applications that not only guarantees deadlines for synchronous messages, but also dramatically reduces the asynchronous messages' response times. Highly responsive guaranteed service is introduced for alert class asynchronous messages. This algorithmic-based scheduling approach supports a priori determination of system timing correctness for arbitrary synchronous and asynchronous message sets without the costly time-line development, testing and tuning associated with TDM techniques. This capability allows token rings to support dynamic, adaptive and reconfigurable run-time environments. In addition, a new Enhanced Responsiveness Protocol that directly embeds real-time scheduling theory into the media access protocol is developed that offers higher performance than is possible with standard-based methods. These new approaches introduce guaranteed alert class service, and dramatically enhanced asynchronous class message service to LANs. The result is highly responsive real-time rings that can form the backbone of predictable, stable extensible real-time systems.