Home // VALID 2016, The Eighth International Conference on Advances in System Testing and Validation Lifecycle // View article

Anomaly-Detection-Based Failure Prediction in a Core Router System

Authors:
Shi Jin
Zhaobo Zhang
Gang Chen
Krishnendu Chakrabarty
Xinli Gu

Keywords: Anomaly Detection; Failure Prediction; SVM

Abstract:
Prognostic health management is desirable for commercial core router systems to ensure high reliability and rapid error recovery. The effectiveness of prognostic health management depends on whether failures can be accurately predicted with sufficient lead time. However, directly predicting failures from a large amount of historical logs is difficult. We describe the design of an anomaly-detection-based failure prediction approach that first detects anomalies from collected time-series data, and then utilizes these “outliers” to predict system failures. A feature-categorizing-based hybrid anomaly detection is developed to identify a wide range of anomalies. Furthermore, an anomaly analyzer is implemented to remove irrelevant and redundant anomalies. Finally, a Support-Vector-Machine (SVM)-based failurepredictor is developed to predict both categories and lead time of system failures from collected anomalies. Synthetic data generated using a small amount of real data for a commercial telecom system, are used to validate the proposed anomaly detector and failure predictor. The experimental results show that both our anomaly detector and failure predictor achieve better performance than traditional methods.

Pages: 20 to 26

Copyright: Copyright (c) IARIA, 2016

Publication date: August 21, 2016

Published in: conference

ISSN: 2308-4316

ISBN: 978-1-61208-500-5

Location: Rome, Italy

Dates: from August 21, 2016 to August 25, 2016