Response time bounds for event handlers in the priority based functional reactive programming (P-FRP) paradigm

Abstract

Priority-based Functional Reactive Programming (FRP) is a new declarative approach to modeling and building reactive systems. Preempted tasks in P-FRP are aborted and have to restart when higher priority tasks have completed. Unlike the preemptive model of execution, there is no notion of critical instance in P-FRP. Determination of the actual value of Worst-Case Response Time (WCRT) in P-FRP requires evaluation of an exponential number of release scenarios. Such a 'brute-force' computation is expensive and impractical in several situations. In previous work, researchers have presented a polynomial time algorithm which uses fixed point iteration to compute an approximate upper bound on response time. However, this algorithm fails to converge for several task sets and hence is quite limited in use. In this paper, we present a more robust algorithm that computes response time bounds of P-FRP tasks and guarantees a result for all task sets. Experimental results using synthetic task sets of different sizes have also been presented.