Research On Real-time Scheduling Algorithm In Intermittent Computing System

With the development of the Internet of Things,low-power devices for Io T are becoming more and more common in our lives,such as wearable devices like smart watches and long-term environmental monitoring facilities.Previously,most of these devices relied on battery power.However,with the increasing requirements for environmental protection and longer running time,intermittent systems that rely on energy harvesting devices as power supply have bocome a research hotspot.The instability of the energy harvesting device causes the system to lose power due to insufficient energy,which gives it the characteristic of intermittent system.In order to provide energy support for intermittent system,this paper studies mainstream energy harvesting platforms and ralated technologies.The built energy harvesting paltform was used to collect dynamic and static energy trajectories for subsequent algorithm design.In view of the different ways of intermittent systems to deal with power failure,this paper builds the Alpaca system based on the task programming model,and measures the execution time of benchmarks on the system.This paper will design scheduling algorithms so that more jobs can be scheduled while meeting real-time requirements.In the system based on task programming model,this paper disigns the scheduling algorithms based on this system from two perspectives of integer programming and heuristic algorithm.The integer programming algorithm constructs a nonlinear integer programming model of the intermittent system scheduling problem,and calls the solver to obtain a scheduling result that conforms to the model.Heuristic algorithms include offline scheduling algorithm and online scheduling algorithm,which find better feasible solution by setting priority and greedy ideas,and so on.In the system based on the checkpoint model,this paper constructs an integer programming problem model based on the checkpoint system,and calls the solver to obtain the scheduling result.In order to better conduct real-time scheduling research,this paper generates task sets by designing intermittent system real-time scheduling simulator,calling scheduling algorithms and simulating scheduling results,so as to better understand the execution process of jobs in real systems.