Stochastic Rate Control for Scalable VBR Video Streaming over Wireless Networks

Video streaming over wireless links is a challenging problem due to both the unreliable, time-varying nature of the wireless channel and the stringent delivery requirements of media traffic. Layered encoded video can be used to improve the system performance by adapting the sending rate for different video frame layers to the varying network and playout situations. In this paper, we study the adaptive control of sending rates for both the base layer and enhancement layer based on feedback information from the wireless receiver. We formulate the problem in a framework of Markov decision processes to minimize a weighted sum of video quality and playout continuity degradation. In order to decrease the computation complexity, we then develop an online greedy algorithm, which only considers the current control time period. Simulation results show that the propose adaptive rate control provides significantly improved video quality and playout smoothness. Furthermore, when rate control is not performed very frequently, the greedy algorithm achieves a video distortion rate nearly matching that of the ideal optimal Dynamic Programming policy.