Why Video Needs Middleware draft-klingenstein-why-vid-needs-mid-00.txt Ken Klingenstein kjk@internet2.edu July 9, 2001 Video information and communication, in many forms, is an extremely compelling modality. Three uses are of particular interest to our community: videoconferencing, video on demand, and multimedia objects and records. Videoconferencing is dramatically changing how we do collaboration in higher education. Whether scheduled for classes and meetings or persistently available for drop-in "water cooler" conversations, videoconferencing is applicable to many different teaching, learning, research, and clinical applications. H.323 is the current vendor supported technology that allows for multi-vendor interoperability, and newer forms of desktop conferencing are emerging. Collaboration environments such as the Access Grid (an environment for persistent video presence ) and VRVS (coming out of the high energy physics community, where it is used for production services ) are becoming more widespread and approaching critical mass for externalities. Video on demand covers a range of new services and capabilities. They range from delivery of instructional materials to students in dorms to streaming sports events to distant alumni. Security video cameras... Complex data structures that include video are being developed for strategic markets. In medicine, medical records may include video interviews with doctors. In humanities, virtual events integrate video streams from a... Underneath these areas like critical common tools sets. Some, such as MBone and H.323 are relatively established, while new tools such as MPEG-7 and ??? The environment appears ready for action. For example, MBone code is now being put into OpenMash software base (VRVS and Access Grid are also now working with OpenMash group, which leads to optimism that we will converge on a single source base for the MBone tools). . The international energy being brought to the VidMid group is also an indicator of widespread interest. However, there is one constant among the variety of different and compelling video applications described above. That constant is that if the applications are to scale from their disjoint nature and limited usage today into the ubiquitous and consistent service that we envision, they need to be linked into an emerging middleware infrastructure that will provide the applications with management, information and security. Almost all video applications will require scalable, navigable *identifiers* and *authentication services* associated with those identifiers. Identifiers typically have two aspects: a set of technology issues (how are they formed, how are they found, in what community are they unique, etc.) and a set of policy issues (who can get them, can they be reassigned, can a subject ever switch its identifier, etc.). Videoconferencing is the most obvious application that would welcome a user-friendly, portable, global identifier that could be used to easily and securely initiate and manage sessions. Video on demand needs such an identifier to enable accounting and pay-per-view. Complex objects need names for the video components that are consistent and extensible. For some identifiers, authentication services can be required to vouch that the bearer of the identifier has proven that fact. Billing needs to have assurance that identity has been proven. Authentication is also helpful in the setup of a video-call; a recipient wants to know that the call originator before taking a call. Authentication, through digital signatures and timestamps, are needed by video objects. Video applications will require a mesh of information and management services that can be effectively and securely traversed across the global Internet. *Directories* provide that service. Video attributes of interest could include a person's video address for videoconferencing, or their access permissions for video on demand. Directories can also describe video objects, where attributes might include encoding information, intellectual property owners, etc. For directories to work, there must be standards in schema design so that attributes have consistent meaning and syntax and are kept in locations that can be automatically discovered. The specific work that the VidMid group seeks to do includes: