Mobile Serverless Video Communication

Abstract

Voice and video conferencing have well established as regular communication services within the wired Internet. Facing the paradigm of ubiquitous computing and mobile communication, they are on the spot to be launched within a wireless Internet infrastructure. Following an 802.11, 802.16 or 3G standard, wireless networks provide enough bandwidth to support data intensive communication services such as videoconferencing. The vision of nomadic users at roaming devices performing synchronous communication, such as voice or videoconferencing over IP (VoIP/VCoIP), is around, but raises new challenges for the Internet infrastructure. In conferencing scenarios addressability raises the first major issues. To globally call a device, a routable IP address must be in use. On a large scale such address space is only provided by IPv6. To identify a communication partner’s current device, a supplementary global user locating scheme is needed. In wireless infrastructures, where users share limited bandwidth from a restricted frequency space, multicasting is needed to enable group conferencing compliant to transmission resources and without placing the burden of dedicated group-server infrastructure. At the same time, synchronous real-time applications s. a. VoIP and VCoIP place new demands on the quality of IP mobility services: Packet loss, delay and delay variation (jitter) in a constant bit rate scenario need careful simultaneous control. A spoken syllable is about the payload of 100 ms continuous voice traffic. Each individual occurrence of packet loss above 1 %, latencies over 100 ms or jitter exceeding 50 ms will clearly alienate or even distract the user. Audio and visual streams in video conferencing additionally require tight synchronization. Inter-stream latencies should remain below 30 ms for audio arriving ahead, 40 ms for audio being behind. While uni-directional distribution may compensate quality deficits by buffers, available techniques of hiding packet loss at the cost of delay and jitter or vice versa are of limited use in conferencing. Their requirements impose strong challenges on a mobile Internet scenario. Challenges are even tightened by multicast-based group communication, since in conferencing scenarios each member commonly operates as receiver and as sender. Real-time requirements consequently are a major driving force for the development of a seamless mobile Internet layer. In concordance with communication capabilities video coding techniques have evolved, as well. The latest standard for video coding H.264/AVC (ITU H.264, 2005), although designed as a generic standard, is predestined for applications like mobile video communications (Stockhammer, Hannuksela & Wiegand, 2003). Besides enhanced compression efficiency it delivers also network friendly video representation for interactive (video telephony) and non-interactive applications (broadcast, streaming, storage, and video on demand). H.264/AVC provides gains in compression efficiency of up to 50% over a wide range of bit rates and video resolutions compared to previous standards. While H.264/AVC decoding software has been successfully deployed on handhelds, high computational complexity still prevents pure software encoders in current mobile systems. There are however also fast hardware implementations available (see a list in Wikipedia, H.264, 2006). Next generation codecs like Scalable Video Coding (SVC) are already in a design state (Reichel, Schwarz, & Wien, 2005, Schwarz et.al., 2004). The main new feature, scalability, addresses schemes for delivery of video to diverse clients over heterogeneous networks, particularly in scenarios where the downstream conditions are not known in advance. The basic idea is that one encoded stream can serve networks with varying bandwidths or clients with different display resolutions or systems with different storage resources, which is an obvious advantage in heterogeneous networks prevalent in mobile applications.