Network Quality Adaptive Video Transmission

Steady improvements in bandwidth offerings over wireline and most recently even wireless networks has helped the emergence of so-called Over the top (OTT) services, that is, quality-sensitive applications which adapt to device and network conditions to deliver a service with a suitable quality for its users. Unlike traditional multimedia deployed over quality-enabled networks, OTT services cannot rely on guaranteed quality levels from the network, nor on feedback from the network on achievable quality. They depend on their own feedback to report on the quality received and infer which mishaps may be occurring at the network level. We present here an algorithm which manages the quality level of a live video stream using a standard feedback mechanism to adapt throughput to varying network conditions. It uses redundancy to both protect traffic from losses but also as a form of safety margin to both predict available throughput and isolate random fluctuations. We also show the performance of the algorithm compared to alternative solutions.