Accelerating Task Completion in IIoT Based Smart Factory Through Optimal Restart

Industrial Internet of Things is a platform designed to support a variety of emerging applications in manufacturing, which are delay sensitive and require ultra reliability. The application tasks are relied on massive distributed sensors and edge computation nodes. These computation nodes are placed close to sensors to provide the low latency response and parallel computation service. The edge node is a device with restricted computation power, it connected to sensors or a central system via the wireless network. As the wireless network often becomes unstable, the reliability of data transmission might be impaired by the unstable connection. In this paper, we investigated a restart mechanism to mitigate these effects. In the presence of unreliable wireless connection, once the waiting time for the completion of a task exceeds a given threshold, exploiting the resource of another node can accelerate the task completion. We analysed the restart scheme with a representative system that consists of two working nodes and one back up node. We mathematically defined the optimization problem of the selection of the optimal timeout. A gradient based algorithm is proposed to solve the optimization problem. The performance comparison between our algorithm and two classic algorithms, AdaDelta and CMA-ES, demonstrated that our algorithm can identify the global optimum in the shortest time.