Lightweight Passive Monitoring for Soft Anomaly Classification in Wired Networks on Resource Constrained Microcontrollers

Prabin Dhakal, Francesco Picariello, Basanta Joshi, Nanda Bikram Adhikari
Abstract:
This paper presents a method for detecting soft anomalies in Ethernet cables using time domain signal analysis and machine learning. A dataset consisting of oscilloscope captured signals from a CAT5e cable using passive measurement technique under four controlled scenarios was used. To reduce data dimensionality while preserving statistical characteristics, a histogram-based feature extraction process was applied prior to classification. Classification was performed using Decision Tree, Random Forest, and Support Vector Machine (SVM) algorithms under various downsampling rates. Models performances were good at generalizing and predicting the classes using those features. The results demonstrate that histogram features are effective in distinguishing between different anomaly types, even at significantly reduced sampling rates showing potential for real time implementation on low powered microcontroller platforms.
Download:
IMEKO-TC6-2025-070.pdf
DOI:
10.21014/tc6-2025.070
Event details
IMEKO TC:
TC6
Event name:
TC6 M4Dconf2025
Title:

2025 IMEKO TC-6 International Conference on Metrology and Digital Transformation

Place:
Benevento, ITALY
Time:
03 September 2025 - 05 September 2025