Dynamic Priority Assignment Technique for Periodic Applications with (m,k)-firm Deadlines


Peter Gschirr


In distributed systems as such automotive, an application is often partitioned into multiple tasks and they communicate over a shared bus systems by exchanging signals or messages. Traditionally, the messages are associated with a deadline in such settings. That is, they must reach their destination within a specific time-duration in all occasions. Such hard deadline constraints are verified and met using the techniques originated from schedulability analysis. In contrast to such requirements, many applications such as feedback control are able to guarantee proper functionality even if some of the associated messages (i.e., feedback signals) miss their deadlines. In literature, such deadline constrains are referred to as (m,k)-firm deadlines, i.e., k out of m message must meet their deadlines with k lesser or equal to m. Such requirements are different from soft real-time requirement because of the stringent performance constrains of the associated application (e.g., feedback control). In this work, we presented a dynamic priority assignment technique which guarantees to meet such (m,k)-firm requirements. We demonstrated the applicability of the proposed scheme in the context of FlexRay based ECU network