Joint request routing and video adaptation in collaborative VoD systems

Collaborative content storing and requesting are emerging solutions to Video-on-Demand (VoD) systems challenged by the explosive growth of online video demand. By exploiting the collaboration among the serving nodes in a VoD system, the new solutions dramatically increase the system capacity. In this work, we focus on the in-system collaborative request routing problem. Together with the video-adaptation techniques, we optimize the aggregate utility of end users while keeping the maximum link congestion under a given upper bound. We prove that the formulated problem is NP-hard, and solve it in two phases: a rate-selection phase and a fixed-rate routing phase. Afterwards, we convert the raw output of the proposed algorithm to a hop-by-hop routing-decision protocol which can be easily implemented in most existing routers. Numerical results show that our solution to the fixed-rate routing problem supports up to 60% more traffic than existing shortest-path algorithm. With joint video adaptation, our algorithm for optimizing utility outperforms the simple greedy method up to 18%.