|
Research on network cloud equipment anomaly and root cause analysis
|
Authors: Dandan Zou, Jianbing Ding, Xidong Wang, Xiaozhou Ye, Ye Ouyang
Status: Final
Date of publication: 16 May 2022
Published in: ITU Journal on Future and Evolving Technologies, Volume 3 (2022), Issue 2, Pages 89-97
Article DOI : https://doi.org/10.52953/TVLO2995
|
Abstract:
With the development of 5G communication technology, a cloud computing system has become a trend. However, with the expansion of the scale of deployment and the increase in framework complexity, ensuring the security and stability of cloud-based systems has become a serious challenge. In a real business environment, existing algorithms are powerless in the face of current problems, such as complex types of abnormal logs, inaccurate time information, and the lack of key information. This paper proposes network cloud equipment anomaly detection and a root cause analysis scheme based on large-scale logs in distributed cluster systems. The scheme uses unsupervised integrated learning, keyword search, and root cause generalization to analyze logs, accurately find anomalies and locate root causes. The F1 score in log anomaly detection is 0.962, and the accuracy in root cause location of anomalies is 0.849. In the ITU AI/ML in 5G Challenge 2021, the solution got the highest final score 93.904 in the China Mobile problem statement of network cloud equipment anomaly and root cause analysis. Furthermore, the scheme has been deployed on China Mobile's 5G network management system and achieves the detection and location of anomalies under intelligent operation. |
Keywords: Artificial intelligence for IT operations, ensemble learning, keyword search, log anomaly detection, network cloud equipment, root cause analysis
Rights: © International Telecommunication Union, available under the CC BY-NC-ND 3.0 IGO license.
|
|
| ITEM DETAIL | ARTICLE | PRICE | |
|---|
ENGLISH 
Full article (PDF) |
|
| 0
| Free of charge | DOWNLOAD |
|
|
| |
