The Effect of Channel Variations on Optimal Download Policies in SVC video delivery with buffer restrictions

In this work, we address the problem of scalable video delivery to mobile users under varying buffering constraints on the user side to determine how channel variations affect the optimal delivery policy. Typically, buffering constraints dictate how much video data the consumer may cache on a local disk ahead of the playback instant, and have a marked impact on the optimal policy for retrieving content. Using a semi-Markov decision process (SMDP) we determine optimal video delivery policies for different levels of buffer limits in different wireless environments for a single user scenario. The outcome of the SMDP shows how the optimal download policy depends on the buffer constraints and the temporal correlation of the wireless channel. We apply a decision tree classifier to the output of the SMDP to derive simple approximate policies for different scenarios. The results indicate that the occupancy of the download buffer is in general more important in the decision making than the current state of the wireless channel. Furthermore, we show that the optimal policy is more conservative in slowly varying, and more greedy in fast changing channels.