2020 European Conference on Optical Communications (ECOC) 2020
DOI: 10.1109/ecoc48923.2020.9333141
|View full text |Cite
|
Sign up to set email alerts
|

Root Cause Analysis for Autonomous Optical Networks: A Physical Layer Security Use Case

Abstract: To support secure and reliable operation of optical networks, we propose a framework for autonomous anomaly detection, root cause analysis and visualization of the anomaly impact on optical signal parameters. Verification on experimental physical layer security data reveals important properties of different attack profiles.

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

0
8
0

Year Published

2021
2021
2023
2023

Publication Types

Select...
3
3
1

Relationship

3
4

Authors

Journals

citations
Cited by 8 publications
(8 citation statements)
references
References 6 publications
0
8
0
Order By: Relevance
“…Then, we propose an Unsupervised Learning (UL)-based RCA algorithm named DB-RCA that does not require prior knowledge on the anomalies caused by physical-layer attacks. The algorithm, which extends our preliminary study in [12], is validated on an experimental physical-layer security dataset. Our performance assessment includes the RCA outputs of eXtreme Gradient Boosting (XGBoost), a known SL model that enables RCA, in addition to our proposed DB-RCA.…”
Section: Introductionmentioning
confidence: 87%
See 2 more Smart Citations
“…Then, we propose an Unsupervised Learning (UL)-based RCA algorithm named DB-RCA that does not require prior knowledge on the anomalies caused by physical-layer attacks. The algorithm, which extends our preliminary study in [12], is validated on an experimental physical-layer security dataset. Our performance assessment includes the RCA outputs of eXtreme Gradient Boosting (XGBoost), a known SL model that enables RCA, in addition to our proposed DB-RCA.…”
Section: Introductionmentioning
confidence: 87%
“…• We investigate RCA approaches and establish an RCA framework applicable to the security assessment task. • We propose a UL-based DB-RCA algorithm, extending the initial study in [12]. • We validate its applicability to autonomous optical network security management using an experimental physical-layer security dataset.…”
Section: Introductionmentioning
confidence: 95%
See 1 more Smart Citation
“…This would allow the security managers to understand why an ML model attributes a particular setting to an attack and which trends in various OPM parameters triggered such a response. To this end, a Root Cause Analysis (RCA) framework was proposed that analyzes the clusters of OPM data constructed by a UL algorithm [10]. When an anomaly is detected, the framework computes the difference between the average value of each feature in the anomalous samples and the average value of the same feature in the closest normal cluster.…”
Section: Achieving Interpretability Of ML Models' Outputsmentioning
confidence: 99%
“…By analyzing the output of the SL-and UL-based algorithms we can assert that, although significantly different, the outputs of both approaches can provide meaningful insight when investigating the root cause of an attack. The contributions of this paper, and extensions with respect to [12] and [9], can be summarized as follows:…”
Section: Introductionmentioning
confidence: 99%