Task scheduling for real-time multi-processor simulations

Abstract

Scheduling of tasks onto multi-processors is an increasingly important problem in the simulation of avionics systems. The problem is difficult due to the many hard real-time constraints imposed on the schedule in the form of processor frame-time limits and latency requirements. In this paper, we present a solution to this real-time scheduling problem using simulated annealing techniques. The running time of the algorithm is fast enough for it to be applied in the rapid reconfiguration of simulation test benches in use at Boeing Flight Systems Laboratory. Its efficacy is demonstrated using an example with 60 tasks communicating through 1800 common blocks and scheduled onto 6 processors under 4 latency constraints which achieved a utilization factor over 95%. Such an example can be scheduled in approximately 35 minutes of CPU time on an HP-Apollo 425 workstation.<>