Description
Keynote Speaker at RTNS 2012. Title: Getting ones priorities right Abstract: It is now 40 years since the first seminal work on priority assignment for real-time systems that use fixed priority scheduling. Since then, huge progress has been made in real-time scheduling with more complex task models and schedulability analyses developed to better represent and analyse real systems. In this talk, we will take a tour of priority assignment through the ages; a “short history of Pri”. We’ll look at simple task models where deadline monotonic priority assignment is optimal and see how departures from the model break this optimality. We’ll look at Audsley’s algorithm for optimal priority assignment (OPA), including the rules for when we can and can’t use it – as well as a catalogue of situations where it’s useful. We’ll see how this algorithm can be extended to form robust priority assignments, and how they can be used to define priority orderings even when we only have partial information about the system. We’ll look at what to do when we can’t use the OPA algorithm – with the aim of avoiding the need to check all possible priority orderings. We’ll look at priority orderings for systems with execution times described by random variables, and examine systems where tasks have two priorities and ask what can be done in that case. Finally, we’ll recap on why priority assignment is so important and discuss some open problems in this area.| Period | 8 Nov 2012 |
|---|---|
| Event title | 20th International conference on Real-Time and Network Systems (RTNS) |
| Event type | Conference |