The Impact of Network Applications (Web, FTP and E-mail) and Network Profiles on the Self-Similar Parameter

Self-similarity is an important characteristic of modern network traffic, and an accurate traffic description must take this characteristic into account. This work analyzes traffic's self-similar behavior for the three most expressive application protocols used in IP networks (HTTP, FTP and SMTP) by varying the user profiles (in terms of the transferred file size and file inter-arrival times) and the network characteristics (with respect to the transfer delay and packages loss rate). The simulation results show that the intrinsic characteristics of application protocols, of the TCP transport protocol, and of user and network profiles affect the traffic self-similar characteristics in distinct ways. Experimental results also reveal that different quantities of contribution from different application protocols in the traffic load can also change traffic's self-similar characteristics.