Zajem in analiza metapodatkov v omrežju BitTorrent s pomočjo porazdeljene razpršene tabele

Šijanec, Anton Luka (avtor), Ciglarič, Mojca (mentor)

Diplomsko delo predstavi protokol BitTorrent in njegov podporni sistem DHT (porazdeljena razpršena tabela) in se osredotoči na uporabo slednjega za pridobivanje metapodatkov iz omrežja BitTorrent. Za pridobivanje smo spisali namenski program, ki z normalnim sodelovanjem v omrežju DHT na podlagi poizvedb drugih uporabnikov pridobiva zgoščene vrednosti metapodatkov (imena datotek in njihove velikosti), ki jih nato od soležnikov tudi prenese. Na podlagi štirih zajemov v letih 2023, 2024 in 2026 (skupaj okoli milijon in pol torrentov) analiziramo stanje omrežja. Ugotavljamo, da je najpopularnejši odjemalec qBittorrent, sledi μTorrent, katerega popularnost pada. Vodilne končnice imen datotek so po vrsti mp4, mkv, avi, mp3, zip in flac. Sodeč po imenih datotek je najpopularnejša ločljivost videovsebin 1080p, popularnost 480p pada, 2160p pa raste. Delež torrentov, prenesenih prek IPv6, raste, leta 2026 smo po IPv6 prenesli 17,5% torrentov. Več kot polovico torrentov iz Mehike, Brazilije in Romunije smo prenesli preko IPv6, od slovenskih naslovov pa manj kot 5%. Klasifikacija na kategorije z jezikovnim modelom je pokazala, da piratsko vsebino vsebuje 93% torrentov, da je 28,9% torrentov pornografija, sledijo TV oddaje oz. serije s 15,2%, filmi s 14,7% in glasba z 11,5%.

• Jezik: Slovenski jezik
• Ključne besede: BitTorrent, DHT, P2P, Kademlia
• Vrsta gradiva: Diplomsko delo/naloga
• Tipologija: 2.11 - Diplomsko delo
• Organizacija: FRI - Fakulteta za računalništvo in informatiko
• Leto izida: 2026
• PID: 20.500.12556/RUL-185122
• COBISS.SI-ID: 286542851
• Datum objave v RUL: 23.07.2026

Sekundarni jezik

• Jezik: Angleški jezik
• Naslov: Acquisition and analysis of BitTorrent network metadata using DHT

Izvleček

The thesis introduces the BitTorrent protocol and its supporting protocol, the DHT (distributed hash table), focusing on using DHT to acquire metadata from the BitTorrent network. We developed specialised software that records queries sent by other users’ clients to our observation node in the DHT network. Such queries contain infohashes which are used to download metadata (names and sizes of files) from peers. From four independent acquisitions in the years 2023, 2024 and 2026, containing around a million and a half torrents, we infer the state of the BitTorrent network. The most popular client is qBittorrent, followed by μTorrent, falling in popularity. Leading filename extensions are in order mp4, mkv, avi, mp3, zip and flac. The most popular video resolution according to filenames is 1080p, the popularity of 480p is falling and of 2160p is rising. The ratio of torrents received via IPv6 is rising; in 2026 we received 17.5% of torrents via IPv6. More than half of torrents from Mexico, Brazil and Romania were received via IPv6, but less than 5% of torrents received from Slovenian IP addresses were received via IPv6. Classification with a large language model shows that 93% of torrents are for the purposes of piracy, that 28.9% of torrents are pornography, followed by TV shows with 15.2%, movies with 14.7% and music with 11.5%.

• Ključne besede: BitTorrent, DHT, P2P, Kademlia