We study the dynamics of a large popular commercial Web site designed specifically for users who access it via their cell-phones and PDAs. Unlike most previous Web studies that have analyzed accesses seen by proxies and servers from clients connected via the wired network, we focus primarily on client accesses made over wireless channels and made for downloading content on small devices for offline browsing. We carry out user-behavior analysis as users authenticate themselves before accessing and then every access is logged with a unique user identifier. Using browser traces gathered over a period of 12 days, we perform detailed content analysis, document popularity analysis and server load analysis. We answer questions like what sorts of content wireless users are most interested in, when and how much load they put on the servers, and how much time they spend on the channel while accessing the Web wirelessly. We discuss the implications of our findings for techniques such as query caching, serverscheduling, channel use and TCP optimization.
展开▼