On the Energy Consumption Forecasting of Data Centers Based on Weather Conditions: Remote Sensing and Machine Learning Approach

The energy consumption of Data Centers (DCs) is a very important figure for the telecommunications operators, not only in terms of cost, but also in terms of operational reliability. A reliable weather forecast would result in a more efficient management of the available energy and would make it easier to take advantage of the modern types of power-grid based on renewable energy resources. In this paper, we exploit the capabilities provided by the FIESTA-IoT platform in order to investigate the correlation between the weather conditions and the energy consumption in DCs. Then, by using multi-variable linear regression process we model this correlation between the energy consumption and the dominant weather condition parameters in order to effectively forecast the energy consumption based on the weather forecast. This procedure could be part of a wider resources optimization process in the core network towards an end-to-end (e2e) access/core network optimization of resources utilization. We have validated our results through live measurements from the RealDC testbed. Results from our proposed approach indicate that forecasting of energy consumption based on weather conditions could help not only DC operators in managing their cooling systems and power usage, but also electricity companies in optimizing their power distribution systems.