Receiver-Driven Adaptive Layer Switching Algorithm For Scalable Video Streaming Over Wireless Networks

This paper considers a receiver-driven adaptive rate control scheme for scalable video streaming over wireless networks by jointly utilizing the channel condition and the receiver buffer state. In order to adjust the source video layers based on the feedback received from the end-user, we define buffer overflow probability (BOP) and buffer underflow probability (BUP) as metrics to characterize the degree of matching between the source video bitrate and the channel throughput. Thus, the problem is formulated as maximizing the attainable video quality while keeping BOP and BUP in an acceptable range. Based on the large deviation principles, we derive a probability estimation model for BOP and BUP. Relying on this estimation model, an online adaptive layer-switching algorithm is designed in order to adapt the video bitrate to the available channel throughput. Simulations are conducted with video trace and the results illustrate that the proposed policy is capable of accommodating different channel qualities without any prior knowledge.