Peer Assisted Parallel Downloading System
Akshat Shetty, Simran Mhatre, Nisheet Sinvhal, Kailas Devadkar · 2019
Popular Peer to Peer (P2P) file-sharing applications like BitTorrent are using segmented file transfer for faster downloads. Peers request chunks of a file from each other and thus retrieve these chunks using cryptographic hashes. This concept of simultaneously downloading different portions of the computer file from either multiple servers or a from a single server is a time tested method of improving downloading speeds. This paper presents a peer-assisted parallel downloading system which uses the concept of segmented file transfer to decrease the download time and also aims to be truly decentralised. The proposed system has peers, that can offer high data transfer rates and are in close proximity to the peer interested in downloading the file, themselves download chunks of the file and stream it to the requestor. This allows the utilization of unused bandwidth while reducing the download time. The use of a custom scoring logic which penalises free riders in the network to contribute towards assistance helps maintain balance in the network while reducing the overall time for download for peers with low data transfer rates. Tests show the overall download time for files can be effectively reduced by 15% for peers with poor data transfer rates.