Key points are not available for this paper at this time.
Data centers are extremely important facilities that contain core business information and applications, but the energy consumption is an inevitable issue, so the thermal distribution monitoring is critical for data centers reliable and stable operation. Thermal information enables monitoring and autonomic thermal management in large data centers. Recent approaches that employed a mobile robot or a thermal camera network can already collect the physical positions of hot spots. However, most of the features extracted from thermal images, such as shape and gradient, may be affected by changes in the position and direction of the thermal camera. In this paper, a thermal distribution monitoring method is proposed to extract the thermal features of a hot spot or a server in a container data center. In the first stage of our method, thermal and visual images are registered based on texture characteristics in images that are obtained in data centers. Then, the thermal distribution of each server is standardized. The features of the hot spot or server extracted from the standard distribution can diminish the impact of camera position and direction. The experimental results verified the efficient and the advantages of this method.
Liu et al. (Sat,) studied this question.