Parallelization And WCRT Analysis Of Space Debris Detector-debie

Most of the traditional embedded systems are based on single-core processor architecture.With the development of low-power multi-core technology, the traditional single-core processors are being replaced by multi-core processors.DEBIE is a program of space debris detector developed by Patria Finavitec and UniSpace Kent.It is urgently needed to be migrated to low-power multi-core processor. However, because of the conflict and interference of shared resources, this work has two mainly technical difficulties. The first is how to make parallazation of hard real-time tasks effective; the second is how to guarantee that the hard real-time tasks must terminate in the respective period of time, which is the problem of the WCET(Worst Case Execution Time) analysis of the program, thus, can ensure the safety of hard real-time tasks schedulingIn this paper, three works are completed as following: First, we decompose DEBIE benchmark into several functional modules and analyze their relationship. Then we make them run in parallel multi-thread manner and get its MSC(Message Sequence Chart). Second, for the task modules of DEBIE in low-power multi-core architecture, we find the optimized shared cache partition and propose the compute method of its WCRT(Worst-Case Response Time). Third, we validate the effectiveness of parallelization approach in this article by experiment. The experimental result shows that the speedup of parallel DEBIE on two different machines I7 and Q6600 are 1.26 and 1.33 and the maximum worst-case speedup is 2.03.