Job scheduling in a distributed computing system on a chip with power consumption minimization

Abstract

Scheduling of computing operations takes an important place in the process of distributed information processing and control systems design, especially in conditions of limited energy resources of the system. This becomes especially important for computers located on autonomous carriers, such as unmanned aerial vehicles, autonomous underwater vehicles, etc. The energy resources in such systems are limited that leads to high requirements for the energy efficiency of the carrier systems including computing ones. The paper presents the job scheduling method for a distributed computing system on a chip which allows reducing the power consumed by the system. The proposed task scheduling method includes two stages. At the first stage, jobs are assigned with the determination of an energy-efficient architecture of the system characterized by the minimum power consumption. At the second stage, jobs are scheduled taking into account the criterion that minimalizes the average job implementation time. A feature of the problem being solved in this case is the necessity of job scheduling in the system with more than one information output which does not allow applying any of the known scheduling methods to the system. The first stage of the proposed method is implemented by implementation additional processors with a simultaneous decrease in the clock frequency and supply voltage. For the second stage of the method, the job scheduling algorithm is proposed which involves the preliminary construction of a private schedule for each output of the system with further integration of these schedules into the general schedule using a heuristic procedure. The scheduling algorithm functioning is illustrated by an example of a solution for a simple system. The advantage of the proposed heuristic method is the possibility of scheduling calculations, taking into account criteria of the minimum power consumption and the minimum average residence time of a task in the system simultaniously. This makes it possible to increase the energy efficiency of solving problems in distributed computing systems on a chip, which contributes to increasing the autonomy of systems in which they are used in. The proposed algorithm has polynomial complexity, therefore, due to the relative simplicity of the algorithm, it can be used for scheduling and rescheduling jobs in real time for complex systems.

Keywords

• computing scheduling
• real-time distributed computing system
• power consumption reduction
• flow shop scheduling
• energy-efficient computing