We review the design and implementation of UltraGrid, a new high definition video conferencing system, and present some experimental results. UltraGrid was the first system to support gigabit rate high definition interactive video conferencing on commodity systems and networks, and we present measurements to illustrate behavior of production networks subject to such real time traffic. We illustrate the benefits of hybrid IP/provisioned optical networks over best effort IP networks for this class of traffic, and motivate the development of congestion control algorithms for interactive conferencing on best effort IP networks

We describe how the new datagram congestion control protocol (DCCP) can be used as a bearer for the real-time transport protocol (RTP) to provide a congestion controlled basis for networked multimedia applications. This is a step towards deployment of congestion control for such applications, necessary to ensure the future stability of the best-effort network if high-bandwidth streaming and IPTV services are to be deployed outside of closed QoS-managed networks

We discuss the evolution of teleconferencing and networked visualization applications to support 3-dimensional display technologies. The implications of a continuation of Moore's law, coupled with constraints on device clock rate due to power consumption, suggest that future end system and network architectures will become increasingly parallel in nature. Given this, we review trends in programming language design that will ease development of highly concurrent systems, and suggest how network transport protocols might evolve to support them.

We performed end-to-end measurements of UDP/IP flows across an Internet backbone network. Using this data, we characterized the packet reordering processes seen in the network. Our results demonstrate the high prevalence of packet reordering relative to packet loss, and show a strong correlation between packet rate and reordering on the network we studied. We conclude that, given the increased parallelism in modern networks and the demands of high performance applications, new application and protocol designs should treat packet reordering on an equal footing to packet loss, and must be robust and resilient to both in order to achieve high performance

It is well known that congestion control is a key issue for the safe deployment of multimedia applications over IP. We describe our initial experiences implementing TCP-friendly congestion control in a system designed to deliver HDTV content over IP networks. In particular we discuss the effects of packet reordering on the calculated throughput, and highlight the problems this can pose for high-rate applications.

We describe an approach to scaling video conferencing, with the use of active agents. Such agents tileNvideo frames into one, by modification of their respective meta-data and adjustment of their video frame rate if necessary. The spatial tiling agents are located within a network, and participants in the session unicast video to the “closest” agent. The agent then multicast the tiled video to the group of all participants. Results show that spatial tiling increases the ability of the end-user to receive large numbers of video streams and reduces network load both in terms of bandwidth and in packets per second. The result is a significant scaling boost to video conferencing systems.

In this paper we describe our experiments on digital video applications, concentrating on the static and dynamic tradeoffs involved in video playback. Our results were extracted from a controlled series of 272 tests, which we ran in three stages.In the first stage of 120 tests, we used a simple player-monitor tool to evaluate the effects of various static parameters: compression type, frame size, digitized rate, spatial quality and keyframe distribution. The tests were carried out on two Apple Macintosh platforms: at the lower end a Quadra 950, and at the higher end, a Power PC 7100/80. Our quantitative metrics included average playback rate, as well as the rate's variance over one-second intervals.The first set of experiments unveiled several anomalous latencies. To track them down we ran an additional 120 tests, from which we concluded that the video and IO operations were insufficiently tuned to each other.In the next step we attempted to correct this problem, by implementing our own video playback software and accompanying device-level handlers. Our emphasis was on achieving a controlled, deterministic coordination between the various system components. An additional set of 32 experiments were carried out on our platforms, which showed frame-rate increases of up to 325%, with associated reductions in rate variance.