期刊
IEEE TRANSACTIONS ON SUSTAINABLE COMPUTING
卷 7, 期 4, 页码 935-949出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TSUSC.2022.3165016
关键词
IoT visualization; collaborative rendering; resource allocation; task scheduling; situation-aware orchestration
类别
资金
- Discovery Program of Natural Sciences and Engineering Research Council of Canada (NSERC) [RGPIN2018-06254]
- Canada Research Chair Program
This paper proposes a situation-aware orchestration mechanism for resource allocation and task scheduling to address the reliability and sustainability issues of resource collaboration in dynamic IoT systems. By three steps, it achieves objective-driven exploration of collaboration opportunity and precise alignment of resource capacity and task demands through a redundant task scheduling scheme.
Three dimensional rendering enabled IoT visualization provides an immersive operation view across large physical environments by contextually aggregating and visualizing numerous data streams from various systems. The massive resource demand to support real-time and high-quality rendering services can be fulfilled by collaborative rendering among resource-constrained wireless devices. To deliver reliable performance, one main challenge is to achieve reliable and sustainable collaboration in a dynamic IoT system with heterogeneous resource capacity and changing user intent. To overcome such issues, we propose a situation-aware orchestration mechanism of resource allocation and task scheduling. The proposed technique achieves objective-driven exploration of collaboration opportunity among heterogeneous resource by three steps: recognizing dynamic condition of resource and task, including resource reliability and computational demand; understanding the mutual impact of resource condition and task performance in the aspect of energy consumption and latency; precise alignment of resource capacity and task demands via a redundant task scheduling scheme. The proposed task scheduling problem is formulated as an optimization model with the objective of maximizing collaboration utility. A genetic algorithm (GA) with adaptive mating-distance is designed to tackle the NP-hard problem, which improves the optimal solution in simulation by approximately 25% and 30% compared to conventional GA and Greedy algorithm, respectively.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据