Overall System Value Maximization for Resource Constrained Heterogeneous Real-Time Systems

Nowadays, many different applications can be consolidated to the same hardware platform, especially as multi-core platforms become mainstream. These applications may have different user perceived values and may or may not share the same composing tasks. When system resources become scarce and not all applications can be executed, how to select and deploy a subset of these applications to maximize the system value is a challenging problem. In this paper, we first formulate the problem as an integer linear programming problem, and then present a computationally efficient heuristic algorithm, i.e., the Max-Min-Min algorithm, to appropriately select applications and map their tasks to different processing elements such that all the timing requirements of the applications are met and the overall system value is maximized. The experimental results demonstrate that the proposed heuristic approach provides a good balance between the solution quality and the solution computation cost when compared with the solutions obtained by CPLEX solver and by other commonly used heuristic algorithms, respectively.