Visual-Quality-Driven Video Networking over 4G Wireless Broadband

Thesis (Ph.D.)--University of Washington, 2014This dissertation intends to develop an effective and complete cross-layer scheduling and resource allocation algorithm for real-time video camera networks over the OFDMA-based 4G mobile infrastructure. There are two main parts in our research: Visual Quality Driven (VQD) scheduling as well as resource allocation for real-time surveillance video uplink (UL) transmission, and Visual Quality Driven resource assignment for real-time video through an integrated downlink (DL) and UL system over the OFDMA-based networks. In the VQD scheduling for video uplink, we propose an effective real-time video uplink (UL) framework for mobile wireless camera networks over an OFDMA-based infrastructure. Mobile wireless camera stations (CS) unicast their videos in real time to base stations (BS) and then these live video streams are fed to subscribers to facilitate real time video monitoring. Based on the visual quality driven utility function, the target bit rate resulting in the highest possible visual quality is quickly set for each UL video by our algorithm. To optimize the system performance, a real time video packet scheduler and a spectral efficient resource allocation policy are derived. This scheduler is also capable of exploiting the inherent diversity gain due to channel variations. Extensive simulations demonstrate that our proposed method can significantly enhance utility, boost spectral efficiency, and stabilize the video quality. For the integrated system, a comprehensive UL and DL framework for wireless mobile camera networks is proposed. On the UL side, this system collects unicast real-time video streams from wireless mobile camera stations (CSs) and forwards them to the social web platforms having capability to deliver live videos, such as YouTube, Twitter, Facebook, etc. On the DL side, the aggregated video streams are multicast to multiple mobile stations (MSs) to facilitate efficient distribution of these videos in real-time. Since this utility function has also taken the popularity of video contents into consideration, the number of video layers to be uploaded can be quickly determined by our resource allocation algorithm. A previously published opportunistic layered multicasting (OLM) scheduling algorithm is also applied on the DL. Simulation results prove that this proposed novel architecture of ours can significantly enhance the spectral efficiency and the users' satisfaction in both the UL and DL directions