Our paper, “Dynamic encoder profile optimisation for real-time video streaming applications”, has been accepted for publication in the Electronics Letters.
For a real-time video application such as a remote learning session, the main difficulty is to keep the latency low while providing a video stream to viewers with different devices over a variety of bandwidth restricted networks. This Letter proposes an optimisation to dynamically select a subset of predefined different video representations, such that with a limited number of encoders, the largest possible group of viewers receives a decodable video stream at a low latency. This optimisation outperforms the static encoder approach by achieving a fairness of 98.2% with an average client bandwidth utilisation of 84.1% to provide higher quality video.
