Experimental evaluation of transport layer protocols for real-time applications

Abstract

Several transport protocols have been proposed for use in the Internet. The specifications of these protocols are based on the assumption that a general purpose transport protocol can be designed to be used by a variety of applications, including time-constrained applications. A widespread consensus among the researchers is that a general purpose transport protocol is not optimized for any applications, and thus cannot deliver the high performance required by real-time applications such as voice, motion and still video, and large file transfer. This paper investigates the validity of the above claim based on the gathering of empirical data. A generic 'protocol benchmark' is used to assess the real-time performance of a set of known transport protocols, and to analyze their suitability to handle the requirements of real-time applications.<>