A channel variation-aware algorithm for enhanced video streaming quality

The demand for video streaming has been soaring in recent years. However, there is a gap between traffic demand and link capacity due to time-varying throughput fluctuation. This fluctuation results in poor quality video streaming services. The most well-known technique to improve the quality of video streaming is: adaptive video streaming. This technique adjusts the video bit-rate to the time-varying link capacity available to each user, based on the user´s feedback about the link quality. Unfortunately, current quality adaptation algorithms do not incorporate channel time variation. This paper proposes a quality adaptation algorithm that adjusts video streaming quality levels for each user by measuring the channel variation experienced by that user. The algorithm periodically measures the channel throughput and calculates throughput variance between pairs of successive measurements. A quality selection decision is then made by comparing the channel variance to a threshold. This determines whether the quality switching will be aggressive or conservative. Through extensive testing using real datasets, representing multiple video sessions, we show that our proposed algorithm reduces the number of quality switching decisions made while maintaining a high average bitrate, compared to other adaptation schemes.