Predictive video scaling - adapting source coding to early network congestion indicators

A major trend in current Internet communication augments voice conversation with video. Video conferencing over IP (VCoIP) has rapidly spread in the mobile realm, where it faces the problem of heterogeneous, fluctuating network conditions. Scalable video coding enables bandwidth adaptation, but requires guidance by appropriate resource estimators. This work focuses on the analysis and design of an adaptive, bandwidth-aware transmission strategy for real-time multimedia applications like video conferencing. We present an early indicator of network congestion based on jitter variation along with our implementation of a new lightweight sender-based approach to adapt the video codec.