Dynamic Control of Receiver Buffers in Mobile Video Streaming Systems

We propose a novel algorithm to efficiently transmit multiple Variable-Bit-Rate (VBR) video streams from a base station to mobile receivers in wide-area wireless networks. The algorithm multicasts video streams in bursts to save the energy of mobile devices. In addition, the algorithm adaptively controls the buffer levels of mobile devices receiving different video streams according to the bit rate of the video stream being received by each device. Compared to previous algorithms, the new algorithm enables dynamic control of the wireless channel and allows the base station to transmit more video data on time to mobile receivers. This is done by providing finer control over the bandwidth allocation of the wireless channel. The problem of optimizing energy saving has been shown to be NP-Complete. We prove that our algorithm finds a feasible schedule if one exists and always produces a correct schedule even when dropped frames are unavoidable. We analytically bound the gap between the energy saving resulting from our algorithm and the optimal energy saving and show that our results are close to optimal. We analyze the tradeoff between the fine control over bandwidth allocation and energy saving and demonstrate that in practical situations, flexible and finer control of bandwidth allocation will result in significantly lower frame loss rates while achieving higher energy saving. We have implemented the proposed algorithm as well as two other recent algorithms in a mobile video streaming testbed. Our extensive analysis and results demonstrate that the proposed algorithm outperforms the other two algorithms; it results in higher energy saving for mobile devices and fewer dropped video frames.