• Zielgruppen
  • Suche
 

H-probe: an active probing tool for estimating traffic correlations

H-probe  enables researchers to estimate cross-traffic correlations along network paths using active probes. The Python source code is available at GitHub (zip archive)

Overview

H-probe is an online active probing tool for estimating traffic correlations from end-to-end measurements. H-probe does not rely on a receiver as it uses ICMP echo packets. It uses libpcap to capture returning ICMP echo replies. From the timing information H-probe is able to estimate the correlation (covariance) of the cross traffic sharing the end-to-end path with the probing traffic. For Internet aggregate traffic it is known [Leland et al. '94] that it is long range dependent (LRD) with Hurst parameter H. The Hurst parameter can be estimated from the covariance slope that is given by 2H-2. H-probe also implements the aggregate variance method known from [Taqqu et al. '95], which is more robust than the covariance, for estimating H. 

H-probe uses sampling methodology described in: "H-Probe: Estimating Traffic Correlations from Sampling and Active Network Probing", by A. Rizk, Z. Bozakov and M. Fidler. The paper is available at arXiv

H-Probe injects ICMP echo request probes from the sender to the target and captures the corresponding round trip times (RTT) using libpcap. Using the RTTs H-probe estimates the traffic correlations on the end-to-end path. Details of the algorithm are given the paper mentioned above.

 

 

Requirements 

1. Linux operating system 

2. root privileges to use libpcap for packet capture

3. python version 2.6 or 2.7 including the following required python packages: pypcap, numpy, scapy. Additionally the following optional python packages will be used if installed: affinity (pypi), progressbar. To use the live plotting functionality gnuplot is required.