Real-time scheduling in video systems

We consider the problem of mapping video algorithms onto systems of high-performance video signal processors with hard real-time constraints. The mapping problem under consideration is computationally hard due to the many constraints that need to be satisfied. We present a compact mathematical formulation which identifies the decision variables and the constraints that are involved. We demonstrate that the mapping problem is NP-hard in the strong sense. The insight resulting from the mathematical formulation leads to a solution approach in which the problem is decomposed into subproblems called delay management, partitioning, and scheduling. These subproblems are handled with well-known techniques from the literature such as network flow, local search, and constraint satisfaction.