Understanding the Non-Gaussian Nature of Network Traffic

Title: Understanding the Non-Gaussian Nature of Network Traffic

Abstract: This research study aims to investigate the causes of the non-Gaussian nature of network traffic, particularly focusing on the characteristics of greedy flows. The authors analyzed IP flow statistics and hop counts between source and destination nodes, classifying applications by the port number. They found that the main flows contributing to the non-Gaussian nature of network traffic were HTTP flows with relatively small hop counts compared with the average hop counts of all flows.

Main Research Question: What are the main factors contributing to the non-Gaussian nature of network traffic, especially in greedy flows?

Methodology: The authors used data from the MAWI traffic archive, measuring a 100-Mbps link with 18-Mbps CAR. They analyzed one-way US-to-Japan traffic during daily busy hours, removing traces with skewness smaller than 0.4. They divided traces into time blocks and defined a flow for each block based on source and destination IP addresses, source and destination ports, and protocol. Greedy flows were defined as those with more than 20 packets, representing throughput of about 1 Mbps. Hop counts were estimated using the TTL field of IP packets.

Results: The authors found that the main flows contributing to the non-Gaussian nature of network traffic were HTTP flows with relatively small hop counts compared with the average hop counts of all flows.

Implications: This study provides valuable insights into the characteristics of network traffic, particularly in greedy flows. Understanding these factors can help improve network performance and management. The findings also suggest potential improvements in network protocols and algorithms to better handle the non-Gaussian nature of network traffic.

Link to Article: https://arxiv.org/abs/0201004v1 Authors: arXiv ID: 0201004v1