Internet Videoconferencing is a desirable application, but difficult to support as it requires performance guarantees. However, the Internet is made up of best-effort networks and will be for some time to come. Best-effort networks are unable to protect distributed, multimedia applications such as videoconferencing from the effects of congestion. End-to-end adaptive scaling methods allow such applications to operate in the absence of network service guarantees. This thesis extends previous work on two-dimensional media scaling, a method that adapts the bit-rate and packet-rate of media streams. We apply two-dimensional scaling to a commercial codec architecture and evaluate the performance in experiments with live Internet traffic. These experiments show some benefits to incorporating two-dimensional scaling in an Internet videoconferencing application, and indicate directions for further work on two-dimensional adaptive methods.
Click here to get a PostScript copy of this thesis (1.0 MB compressed), or here to get a PDF copy of this thesis (422 KB).
Last revised Sun Dec 7 11:09:08 EST 1997 by jeffay at cs.unc.edu.