A measurement and modeling study of residential Internet user behavior

Understanding residential Internet user behavior from operational networks is critical to the design, evaluation and accounting of the access network and network applications. We present analysis of a data set comprised of a week of RADIUS authentication data taken from a large city Internet Service Provider (ISP). We present basic statistics, including session counts based on time-of-day, session length distribution, session inter-arrival times, user usage time distribution, user traffic distribution, upload and download traffic. We also developed models for these behaviors. Our results provide important data for the simulation and modeling of access network protocols and applications.