A Wavelet-inspired Anomaly Detection Framework for Cloud Platforms

David O'Shea, Vincent C. Emeakaroha, John Pendlebury, Neil Cafferkey, John P. Morrison, Theo Lynn

2016

Abstract

Anomaly detection in Cloud service provisioning platforms is of significant importance, as the presence of anomalies indicates a deviation from normal behaviour, and in turn places the reliability of the distributed Cloud network into question. Existing solutions lack a multi-level approach to anomaly detection in Clouds. This paper presents a wavelet-inspired anomaly detection framework for detecting anomalous behaviours across Cloud layers. It records the evolution of multiple metrics and extracts a two-dimensional spectrogram representing a monitored system’s behaviour. Over two weeks of historical monitoring data were used to train the system to identify healthy behaviour. Anomalies are then characterised as deviations from this expected behaviour. The training technique as well as the pre-processing techniques are highly configurable. Based on a Cloud service deployment use case scenario, the effectiveness of the framework was evaluated by randomly injecting anomalies into the recorded metric data and performing comparison using the resulting spectrograms.

Paper Citation

in Harvard Style

O'Shea D., Emeakaroha V., Pendlebury J., Cafferkey N., Morrison J. and Lynn T. (2016). A Wavelet-inspired Anomaly Detection Framework for Cloud Platforms . In Proceedings of the 6th International Conference on Cloud Computing and Services Science - Volume 1: CLOSER, ISBN 978-989-758-182-3, pages 106-117. DOI: 10.5220/0005913701060117

in Bibtex Style

@conference{closer16,
author={David O'Shea and Vincent C. Emeakaroha and John Pendlebury and Neil Cafferkey and John P. Morrison and Theo Lynn},
title={A Wavelet-inspired Anomaly Detection Framework for Cloud Platforms},
booktitle={Proceedings of the 6th International Conference on Cloud Computing and Services Science - Volume 1: CLOSER,},
year={2016},
pages={106-117},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0005913701060117},
isbn={978-989-758-182-3},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 6th International Conference on Cloud Computing and Services Science - Volume 1: CLOSER,
TI - A Wavelet-inspired Anomaly Detection Framework for Cloud Platforms
SN - 978-989-758-182-3
AU - O'Shea D.
AU - Emeakaroha V.
AU - Pendlebury J.
AU - Cafferkey N.
AU - Morrison J.
AU - Lynn T.
PY - 2016
SP - 106
EP - 117
DO - 10.5220/0005913701060117