Energy Consumption Reduction through Resource Allocation Using Docker
Energy Consumption Reduction through Resource Allocation Using Docker
复制标题
DOI：
10.1109/candarw57323.2022.00042
复制
发表时间：
2022-11
期刊：
影响因子：
--
通讯作者：
Eiji Sugahara;Hiroki Nishikawa;Xiangbo Kong;Hiroyuki Tomiyama
中科院分区：
文献类型：
--
作者：
Eiji Sugahara;Hiroki Nishikawa;Xiangbo Kong;Hiroyuki Tomiyama
安防网络摄像头被广泛应用于各个领域，基于人工智能的图像处理算法提高了这些系统的性能。基于人工智能的算法伴随着较高的能量消耗，这一问题已被相关工作所解决。然而，降低AI能耗仍然是一项根本性的任务。本文提出了一种具有两个应用流的安全摄像头系统。我们使用Docker通过不同的方式为这两个应用分配计算资源，以找到更好的能耗更低的分配方式。实验结果证明了通过分配计算资源来降低能耗的可能性。
Security network cameras are widely used in various fields and AI-based image processing algorithms improve the performance of these systems. AI-based algorithms go hand in hand with high energy consumption and this issue is addressed by related works. However, the reduction of AI energy consumption is still a fundamental task. This paper proposes a security camera system with two application streams. We use Docker to allocate computing resources to these two applications in various ways to find a better allocation that consumes less energy. Experimental results demonstrate the possibility of reducing energy consumption by allocating computing resources.