SCOUT: Data center monitoring system with multiple mobile robots

摘要

Data center management is a critical issue as data center is growing at an enormous rate. As the data centers grow larger, data center management techniques have evolved from a simple alert system into a critical element that enables maximum availability, cost efficiency, and controllability. In this paper, we present a mobile robot-based data center management system called, SCOUT with a cost efficiency and scalability. It is an extended work of our previous work called FalconEye which is a vision-based monitoring system for the maintenance of data centers. Unlike our previous works, we advanced our system by deploying mobile robot platform so that it takes an advantage of monitorability and system controllability. SCOUT is comprised of the following three novel mechanisms: ground integrated NFC tags enabling for the robots to recognize the exact position for a system monitoring; efficient path construction algorithm for a low latency patrol with a fixed number of robots; and the robot path adaptation to maximize the robot utilization. To evaluate the feasibility and practicality of the SCOUT design, we built a prototype of the data center management system with a real robot platform and successfully deployed into iCubeCloud testbed. We expose the possibility of this system by using real monitoring prototype and evaluate algorithms by using SCOUT simulator.