Estimating Method of the Hurst Parameter for IP network traffic

Abstract

Internet traffic and traffic of IP-network exhibits self-similarity and long-range dependence (LRD). In this paper, we propose to adapt the Modified Allan Variance (MAVAR) to estimate the Hurst parameter of LRD traffic traces. The method is applied to pseudo-random self-similar data series, generated with assigned values of self-similarity parameter H. The behavior of MAVAR with most common deterministic signals is also studied. Finally, the technique is applied to a real IP traffic channel, providing the example of MAVAR usage for traffic analysis.