Two major issues arise when designing real-time embedded systems upon multiple processors: the task allocation and the feasibility/schedulability analysis. Clearly, we cannot allocate real-time tasks to processors that are unable to guarantee their deadlines and, hence, these two problems are interdependent and cannot be handled separately. In this thesis, a novel technique is proposed to perform the schedulability analysis of a task set, based on which existing allocation heuristics are extended to consider tasks with arbitrary deadlines. Furthermore, the allocation of communicating tasks is analyzed and new polynomial-time heuristics are proposed to reduce the communication cost among processors.
«
Two major issues arise when designing real-time embedded systems upon multiple processors: the task allocation and the feasibility/schedulability analysis. Clearly, we cannot allocate real-time tasks to processors that are unable to guarantee their deadlines and, hence, these two problems are interdependent and cannot be handled separately. In this thesis, a novel technique is proposed to perform the schedulability analysis of a task set, based on which existing allocation heuristics are extende...
»