MultiResolution Anomaly Detection Method for Long Range Dependent Time Series

TL;DR

This paper introduces a MultiResolution Anomaly Detection (MRAD) method tailored for long-range dependent time series, leveraging multiscale properties to improve network intrusion detection and outperform classical outlier detectors.

Contribution

The paper presents a novel MRAD method with theoretical analysis, including a new formulation showing its increased detection power over single-scale methods.

Findings

MRAD effectively detects anomalies in long-range dependent time series.

Theoretical properties of MRAD are rigorously analyzed.

MRAD outperforms classical outlier detection methods in experiments.

Abstract

Driven by network intrusion detection, we propose a MultiResolution Anomaly Detection (MRAD) method, which effectively utilizes the multiscale properties of Internet features and network anomalies. In this paper, several theoretical properties of the MRAD method are explored. A major new result is the mathematical formulation of the notion that a two-scaled MRAD method has larger power than the average power of the detection method based on the given two scales. Test threshold is also developed. Comparisons between MRAD method and other classical outlier detectors in time series are reported as well.