Abstract:

According to the loginfinitely divisible cascades framework introduced from the analysis of turbulence scale, this thesis analyses the multiscale character of wirelesspacket network traffic gathered in mobile operations' network, sampled with general packet radio service (GPRS)， found that the actual GPRS traffic conforms to the loginfinitely divisible cascades framework; Further more aspects of analysis on wirelesspacket flow data are accomplished based on this method, found that log infinitely divisible cascades (LIDC) is very effective on analysis of wirelesspacket traffic multiscale character. The LIDC can depict the scalecharacter in the fullrange scale of traffic, as well as it can differentiate the valid range within a specific scale. And more, based on this framework, this thesis's analysis shows that the scale of wirelesspacket traffic complies with powerlaw, monoscale and scaleinvariance.

Key words: network traffic analysis, log ilarity, wireless network