Monitoring and Forecasting of Temperature Sensor Data for Large-scale Computing Servers

2014 Fiscal Year Final Research Report

• Project/Area Number: 24500138
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Media informatics/Database
• Research Institution: Kumamoto University (2013-2014); NTT Communication Science Laboratories (2012)
• Principal Investigator: SAKURAI Yasushi 熊本大学, 自然科学研究科, 教授 (30466411)
• Co-Investigator(Kenkyū-buntansha): YOSHIKAWA Masatoshi 京都大学, 情報学研究科, 教授 (30182736)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: データマイニング / 時系列データ / センサデータ

Outline of Final Research Achievements

We developed time-series data mining techniques to predict the future CPU temperature for data-center monitoring. Time-series data analysis is a well-known topic that has attracted huge interest in various research communities (e.g., theory, databases, data mining, networking) for a few decades. The increasing volume of online, time-stamped activity represents a vital new opportunity for data analysis, and the most fundamental requirements are the efficient and effective mining of big time-series data with high variety information. We presented three new
directions for research on time-series analysis, which include: (1) automatic mining, (2) large-scale tensor analysis, (3) non-linear modeling, and we developed efficient and effective data mining techniques.

Free Research Field

時系列データマイニング