MidField: An Adaptive Middleware System for Multipoint Digital Video Communication

Abstract

As broadband networks become more common, using high quality video streams such as DV (Digital Video) or HDV (High-Definition Video) has become increasingly popular in multipoint communications. Today's video communication systems which support high quality digital video formats on broadband networks are used for a variety of purposes (Bak, 2006; Gharai, 2006; Bodecek & Novotny, 2007). Although those systems have many useful functions for multipoint communications, it requires much bandwidth for video streams. A unidirectional DV stream requires about 28.8 Mbps. HDV 720p and HDV 1080i require about 19 Mbps and 25 Mbps respectively. Therefore, when we use high quality digital video communication systems, in many cases we should consider whether each location has enough communication environments to process the video streams or not. Of course, we can’t always use effective format transcoding functions in multipoint communications. As the result, most video formats that are used by current multipoint communication systems are low resolution formats than DV and HDV even if we can use DV/HDV cameras at each end point. If suitable format transcoding functions as defined by RTP (Schulzrinne, 2004) or stream integration functions for large scale video communications (Gharai, 2002) are available on our communication environments, it may be able to transcode the stream format into another suitable stream format or integrate several streams to a mixed stream. If the required communication environments are permanent and the total number of participants is always limited, we may be able to prepare the required transcoding and integration functions into suitable intermediate nodes in advance. However, we can't always use enough bandwidth and CPU power, and our communication environments aren't always permanent, therefore we should consider on demand transcoding functions with relocatable mechanisms. In order to design and implement relocatable transcoding functions for multipoint digital video communications, employing an effective middleware (Ferreira, 2003; Jameela, 2003) on programmable networks (Campbell, 1999) is an effective way. As one of the methods for executable program modules to migrate to anywhere, many mobile agent systems (Guedes, 1997; Ohsuga, 1997; Dong & Seung, 2001; Antonio, 2002; Guo, 2004) also have been proposed. We consider that mobile agents which perform given tasks on suitable nodes are one of important base techniques for constructing flexible and interactive multipoint digital video communication environments.