Agent-based task scheduling in shared context pervasive environment

Abstract

Efficient task scheduling in shared context pervasive environment to collaborate to achieve application requirements, especially real time efficiency, becomes a major issue in recent pervasive applications. In a number of scenarios based on shared context, tasks do have dependency with each other, that we can explore their relationship to implement parallel execution maximization. In this paper, we propose an agent-based solution, called Common Parallelism Approach based on Agents (CPAA) to maximize the task execution in parallel. We further explore the task flow in pervasive applications based on data dependent relationships between tasks. We present CPAA to parallelize three kinds of task flow models based on shared context -- sequence, branch and loop, by agents.