Methods for efficient lossless streaming of interactive applications over a network

Partly due to an increase in available bandwidth, streaming technologies have become increasingly popular. Most current streaming systems focus on distribution of video and music; however, with increasingly cheap hardware and more abundant bandwidth, streaming entire applications, including three dimensional applications, could become feasible in the near future. This thesis proposes and implements a method, herein termed River, of difference-based streaming of frame data.;By buffering rendered frames on both the server and the client side, the server performs iterative pixel-based differences of future frames and reduces transmission costs by only transmitting relevant difference data. Attempts are then employed to reduce bandwidth costs further. Such techniques include popular lossless image compression techniques such as Huffman coding, fast calculation and representation of difference data, and a hashing algorithm to quickly find the most relevant historic frames. Finally, through the use of the software rasterizer contained within the open source Mesa3d graphics package, several intensive test cases are created and streamed using various methods with the resultant timings used as a means of comparison.