A Deadline-Aware Estimation of Distribution Algorithm for Resource Scheduling in Fog Computing Systems

The Internet of Things (IoT) develops rapidly and has produced a large amount of data these years. A range of responsive IoT applications arise and are needed to be processed in a timely manner. Compared with traditional cloud computing system, fog computing is one of the promising solutions of processing the huge amount of local data and decreasing the end-to-end latency. Hard and soft deadlines are assigned to the tasks s. In this work, the resource allocation and task scheduling problem under fog system is considered to minimize total tardiness of the tasks and meet the hard deadlines. A deadline-aware estimation of distributed algorithm (dEDA) with a repair procedure and local search is adopted to determine the task processing order and computing node allocation. The comparative results show that the solution produced by our proposed algorithm performs better than the algorithm without repair procedure or knowledge driven local search. In addition, the performance of our algorithm exceeds significantly the heuristic method on both total tardiness and successful rate metrics. Compared with the existing fog computing resource management algorithm, our algorithm performs much better under most situations.