Measuring and analyzing video downloads in BitTorrent

BitTorrent (BT) is the most used peer-to-peer file sharing system, with the most popular content shared being copyrighted. For the assessment of counter-measures, detailed and complete BT measurements are required. Therefore, this thesis investigates how to collect and analyze complete BT data. Through the comparison of prior BT measurement studies and their resulting data sets, a gap in existing measurements was identified. Specifically, measurements, or data sets, providing multiple samples per hour over weeks while still covering tens of thousands of swarms were missing. Therefore, this thesis makes the following contributions: (a) the design and implementation of a distributed scalable BT measurement system, termed VIOLA, which can collect data sets that fill the identified gap; (b) a data set covering 3 months, 70,000 swarms, and 3 samples per hour, proving that the VIOLA system works as designed and providing the basis for novel insights in BT behavior; (c) a method to transform the measurement data into graphs with different semantics building the basis for quantifying user- and content-centric properties of the BT system.