Research on the Distribution and Self-Similarity Characteristic of End-To-End Network Delay

Abstract

As an important indicator of evaluating network performance, network delay can reflect the transmission performance of the current path and also the service level provided by the opposite terminal host. In the paper, analysis is first conducted on the composition of network delay; besides, through Ping measuring method, Ping measuring network delay experiment is conducted on different destination hosts of local area network (LAN) and wide area network (WAN). Besides, the distribution features of network delay as time changes is analyzed, while the influence of data package (in terms of the size) on network delay will be explored in the paper. After the process, the definition, along with the distinguishing method of the self-similarity process is introduced, and selfsimilarity distinguishing is rendered on network delay through

Self‐similarity based DDoS attack detection using Hurst parameter

“…Typically, arrival pattern of bursty traffic in high‐speed communication network is assumed to follow Poisson distribution. However, from the survey reported in on local and wide area network traffic and from the analysis of other research groups as reported in and , it is evident that the arrival of bursty traffic follows asymptotic self‐similarity model. The basic limitation of Poisson model is that it cannot hold or unable to capture traffic burstiness, which characterizes data traffic (opposite to voice traffic in old telephone systems).…”