Online multi-objective optimization for live video forwarding across video data centers

The proliferation of video surveillance has led to surveillance video forwarding services becoming a basic server in video data centers. End users in diverse locations require live video streams from the IP cameras through the inter-connected video data centers. Consequently, the resource scheduler, which is set up to assign the resources of the video data centers to each arriving end user, is in urgent need of achieving the global optimal resource cost and forwarding delay. In this paper, we propose a multi-objective resource provisioning (MORP) approach to minimize the resource provisioning cost during live video forwarding. Different from existed works, the MORP optimizes the resource provisioning cost from both the resource cost and forwarding delay. Moreover, as an approximate optimal approach, MORP adaptively assigns the proper media servers among video data centers, and connects these media servers together through network connections to provide system scalability and connectivity. Finally, we prove that the computational complexity of our online approach is only O(log(|U|)) (|U| is the number of arrival end users). The comprehensive evaluations show that our approach not only significantly reduces the resource provisioning cost, but also has a considerably shorter computational delay compared to the benchmark approaches.