Online Reinforcement Learning Based HTTP Adaptive Streaming Scheme

DASH is an effective way to improve the Quality of Experience (QoE) in video streaming. However, most of the existing schemes depend on heuristic algorithms, and the learning-based methods that have recently started to appear also have a problem in that their performance deteriorates in a specific environment. In this paper, we propose an adaptive streaming scheme that utilizes online reinforcement learning. The proposed scheme adapts to changes in the client's environment by upgrading the ABR model while performing video streaming when QoE degradation is confirmed. In order to adapt the ABR model to the changing network environment, the neural network model is trained with the state-of-the-art reinforcement learning algorithm. The performance of the proposed scheme is evaluated through simulation-based experiments under various network conditions. Through the experimental results, it is confirmed that the proposed scheme shows better performance than the existing schemes.