Author:
Miao Zicong,Li Weize,Pan Xiaodong
Abstract
AbstractWith the booming of cloud-based digital twin systems, monitoring key performance indicators has become crucial for ensuring system security and reliability. Due to the massive amount of monitoring data generated, data compression is necessary to save data transmission bandwidth and storage space. Although the existing research has proposed compression methods for multivariate time series (MTS), it is still a challenge to guarantee the correlation between data when compressing the MTS. This paper proposes an MTS Collaborative Compression (MTSCC) method based on the two-step compression scheme. First, shape-based clustering is implemented to group the MTS. Afterward, the compressed sensing is optimized to achieve collaborative compression of grouped data. Based on a real-world MTS dataset, the experimental results show that the proposed MTSCC can effectively preserve the complex temporal correlation between indicators while achieving efficient data compression, and the root mean squared error of correlation between the reconstructed and original data is only 0.0489 in the case of 30% compression ratio. Besides, it is verified that using the reconstructed data in the production environment has almost the same performance as using the original data.
Publisher
Springer Science and Business Media LLC
Reference31 articles.
1. Lu Q et al (2020) Digital twin-enabled anomaly detection for built asset monitoring in operation and maintenance. Autom Constr 118:103277
2. Amazon. Major outage hits Amazon Web Services.
https://www.cbsnews.com/news/amazon-web-services-major-outage-many-sites-affected/?intcid=CNM-00-10abd1h. Accessed 15 July 2023
3. He X. Alibaba cloud breakdown affects Hong Kong and Macau. https://www.guancha.cn/economy/2022_12_19_671980.shtml. Accessed 15 July 2023
4. Rabkin A, Katz R (2010) Chukwa: a system for reliable {Large-Scale} log collection. 24th Large Installation System Administration conference (LISA 10)
5. Zhang X et al (2019) Cross-dataset time series anomaly detection for cloud systems. 2019 USENIX Annual Technical Conference (USENIX ATC 19)