Real-Time Big Data Delivery in Wireless Networks: A Case Study on Video Delivery

The huge volume of contents generated from mobile ends has dramatically contributed to big data. Unfortunately, current packet scheduling policies in wireless networks for the underlying big data delivery greatly hinder the utilization of the contents. Specifically, the delivery of real-time big data requires running-time interactions with users, as well as variable bandwidth consumptions to maintain fine user experience. There is no previous work designed for this kind of traffic. To mitigate this gap, a thorough-designed scheduling policy to assign and organize detailed packet transmissions for real-time big data is desired. This paper takes video delivery as a case study, which dominates the current real-time traffic and can be easily extended to other scenarios. We propose a novel scheduling policy, which assigns a proper number of video requests to servers and allocates bandwidth to these requests in a relatively small time scale. It helps with serving more users without compromising user experience of the current ones. We also prove that the scheduling policy has a guaranteed performance on the total number of served requests. Finally, the simulation results demonstrate that our scheduling policy outperforms other state-of-art methods significantly.