Cause of symmetry for traffic volume: Mining and exploring new characteristics of traffic in non-business networks

Measurement methods based on bi-directional flow model are proposed to investigate the symmetric characteristic of the non-business network traffics. The features of protocol, port distribution and flow size are extracted from host-level aggregated traffic to characterize users' interactive behavior and to find the origins of traffic symmetry. The analysis results using actual traffic traces collected from the campus network of Xi'an Jiaotong University show that 1% of the total hosts bear more than 90% of the forward UDP traffic, which results in the symmetry of total traffic. Compared with the traditional flow model, the proposed methods can quickly locate the hosts that are responsible for traffic symmetry, and are more suitable for traffic control and management in non-business networks.