Anomaly Detection in a Data Center with a Reconstruction Method Using a Multi-Autoencoders Model

Early detection of anomalies in data centers is important to reduce downtimes and the costs of periodic maintenance. However, there is little research on this topic and even fewer on the fusion of sensor data for the detection of abnormal events. The goal of this paper is to propose a method for anomaly detection in data centers by combining sensor data (temperature, humidity, power) and deep learning models. The model described in the paper uses one autoencoder per sensor to reconstruct the inputs. The auto-encoders contain Long-Short Term Memory (LSTM) layers and are trained using the normal samples of the relevant sensors selected by correlation analysis. The difference signal between the input and its reconstruction is then used to classify the samples using feature extraction and a random forest classifier. The data measured by the sensors of a data center between January 2019 and May 2020 are used to train the model, while the data between June 2020 and May 2021 are used to assess it. Performances of the model are assessed a posteriori through F1-score by comparing detected anomalies with the data center’s history. The proposed model outperforms the state-of-the-art reconstruction method, which uses only one autoencoder taking multivariate sequences and detects an anomaly with a threshold on the reconstruction error, with an F1-score of 83.60% compared to 24.16%.





References:
[1] Strasbourg datacentre: latest information, Strasbourg datacentre: latest information. https://www.ovh.com/world/news/press/cpl1787.fire-our-strasbourg-site
[2] Average Cost of a Data Center Outage, Data Center Frontier, janv. 25, 2016. https://datacenterfrontier.com/average-cost-of-a-data-center-outage/
[3] L. Decker, D. Leite, L. Giommi, et D. Bonacorsi, « Real-Time Anomaly Detection in Data Centers for Log-based Predictive Maintenance using an Evolving Fuzzy-Rule-Based Approach », 2020 IEEE Int. Conf. Fuzzy Syst. FUZZ-IEEE Fuzzy Syst. FUZZ-IEEE 2020 IEEE Int. Conf. On, p. 1‑8, juill. 2020, doi: 10.1109/FUZZ48607.2020.9177762.
[4] L. Wang et al., « Anomaly monitoring in high-density data centers based on gaussian distribution anomaly detection algorithm », 2020 IEEE Int. Conf. Adv. Electr. Eng. Comput. Appl. AEECA Adv. Electr. Eng. Comput. Appl. AEECA 2020 IEEE Int. Conf. On, p. 836‑841, août 2020, doi: 10.1109/AEECA49918.2020.9213549.
[5] C.-J. Su et S.-F. Huang, « Real-time big data analytics for hard disk drive predictive maintenance », Comput. Electr. Eng., vol. 71, p. 93‑101, oct. 2018, doi: 10.1016/j.compeleceng.2018.07.025.
[6] M. Cakir, M. A. Guvenc, et S. Mistikoglu, « The experimental application of popular machine learning algorithms on predictive maintenance and the design of IIoT based condition monitoring system », Comput. Ind. Eng., vol. 151, janv. 2021, doi: 10.1016/j.cie.2020.106948.
[7] F. l. f. Pereira, D. n. Teixeira, J. p. p. Gomes, et J. c. Machado, « Evaluating One-Class Classifiers for Fault Detection in Hard Disk Drives », 2019 8th Braz. Conf. Intell. Syst. BRACIS Intell. Syst. BRACIS 2019 8th Braz. Conf. BRACIS, p. 586‑591, oct. 2019, doi: 10.1109/BRACIS.2019.00108.
[8] H. V. Dudukcu, M. Taskiran, et N. Kahraman, « LSTM and WaveNet Implementation for Predictive Maintenance of Turbofan Engines », 2020 IEEE 20th Int. Symp. Comput. Intell. Inform. CINTI Comput. Intell. Inform. CINTI 2020 IEEE 20th Int. Symp. On, p. 000151‑000156, nov. 2020, doi: 10.1109/CINTI51262.2020.9305820.
[9] A. P. Hermawan, D.-S. Kim, et J.-M. Lee, « Predictive Maintenance of Aircraft Engine using Deep Learning Technique », 2020 Int. Conf. Inf. Commun. Technol. Converg. ICTC Inf. Commun. Technol. Converg. ICTC 2020 Int. Conf. On, p. 1296‑1298, oct. 2020, doi: 10.1109/ICTC49870.2020.9289466.
[10] S. Hochreiter et J. Schmidhuber, « Long Short-Term Memory », Neural Comput., vol. 9, no 8, p. 1735‑1780, nov. 1997, doi: 10.1162/neco.1997.9.8.1735.
[11] M. Canizo, I. Triguero, A. Conde, et E. Onieva, « Multi-Head CNN-RNN for Multi-Time Series Anomaly Detection: An industrial case study », Neurocomputing, janv. 2019, doi: 10.1016/j.neucom.2019.07.034.
[12] M. Alrifaey, W. h. Lim, et C. k. Ang, « A Novel Deep Learning Framework Based RNN-SAE for Fault Detection of Electrical Gas Generator », IEEE Access Access IEEE, vol. 9, p. 21433‑21442, janv. 2021, doi: 10.1109/ACCESS.2021.3055427.
[13] S. U. Jan, Y. D. Lee, et I. S. Koo, « A distributed sensor-fault detection and diagnosis framework using machine learning », Inf. Sci., vol. 547, p. 777‑796, févr. 2021, doi: 10.1016/j.ins.2020.08.068.
[14] Yassine Bouabdallaoui, Zoubeir Lafhaj, Pascal Yim, Laure Ducoulombier, et Belkacem Bennadji, « Predictive Maintenance in Building Facilities: A Machine Learning-Based Approach », Sensors, vol. 21, no 1044, p. 1044‑1044, févr. 2021, doi: 10.3390/s21041044.
[15] F. L. F. Pereira, I. Castro Chaves, J. P. P. Gomes, et J. C. Machado, « Using Autoencoders for Anomaly Detection in Hard Disk Drives », 2020 Int. Jt. Conf. Neural Netw. IJCNN Neural Netw. IJCNN 2020 Int. Jt. Conf. On, p. 1‑7, juill. 2020, doi: 10.1109/IJCNN48605.2020.9206689.
[16] S. Akcay, A. Atapour-Abarghouei, et T. P. Breckon, « GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training », ArXiv180506725 Cs, nov. 2018, [En ligne]. Disponible sur: http://arxiv.org/abs/1805.06725
[17] H. Ahn, D. Jung, et H.-L. Choi, « Deep Generative Models-Based Anomaly Detection for Spacecraft Control Systems », Sensors, vol. 20, no 7, avr. 2020, doi: 10.3390/s20071991.
[18] D. P. Kingma et J. Ba, « Adam: A Method for Stochastic Optimization », ArXiv14126980 Cs, janv. 2017, [En ligne]. Disponible sur: http://arxiv.org/abs/1412.6980
[19] H. Akaike, « Information Theory and an Extension of the Maximum Likelihood Principle », in Selected Papers of Hirotugu Akaike, E. Parzen, K. Tanabe, et G. Kitagawa, Éd. New York, NY: Springer, 1998, p. 199‑213. doi: 10.1007/978-1-4612-1694-0_15.