So far I have read over the papers listed at http://www.cs.duke.edu/courses/spring07/cps082s/assign/tech_links.html . Based on the data shown there, which I hope to verify with empirical data of my own, the BitTorrent protocol seems to have some promising properties as a potential media distribution system. Bram Cohen's paper from bittorrent.com lists bandwidth requirements of around 1/1000 the total amount transferred. However, the nature of the BitTorrent protocol requires that individual downloaders receive the exact same stream of bits. This poses a problem for many DRM schemes since any individualized protection such as per-user encryption must be applied after downloading, making it easy for moderately skilled hackers to circumvent the DRM simply by copying the data before the encryption is applied. Another issue is that, while BitTorrent can greatly improve performance over centralized downloads for extremely popular content, rarely downloaded files achieve much lower rates than traditional methods, since individual users are unlikely to have comparable bandwidth and uptime to commercial webservers. Two potential methods to ameliorate this problem are to either require users to download files that they do not purchase (which would require some sort of protection to keep them from accessing the data) or having the service provider (such as Apple) have servers dedicated to 'downloading' all the files to ensure availability.
-David, Technical Liaison
No comments:
Post a Comment