Impacts of Inaccurate Information on Resource Allocation for Multi-Core Embedded Systems

Multi-core technologies are widely used in embedded systems. Stochastic resource allocation can guarantee certain quality of the services (QoS). In heterogeneous embedded system resource allocation, execution time distributions of different tasks on cores are predicted before scheduling. The difference between actual execution time and estimated execution time may lead to allocations which are not robust. In this paper, we present an evaluation of impacts of inaccurate information on resource allocation. We propose a systematic measure of the robustness degradation and evaluate how inaccurate probability parameters affect the robustness of resource allocations. Furthermore, we compare the performance of three widely used greedy heuristics when using the inaccurate information.