EAGER: Perceptual-Quality-Aware Video Communication in Wireless Camera Networks

成像硬件和无线通信的最新进展促进了嵌入式摄像头传感器在各种无线成像应用中的部署，例如监视、智能建筑操作、智能交通、远程医疗保健以及消费电子和娱乐。保证摄像机传感器产生的视频以良好的质量交付给最终用户是至关重要的。特别是，保持良好感知质量的需求是由许多以人为中心的成像应用程序驱动的，在这些应用程序中，人类用户依赖于接收到的视频来做出关键决策。然而，许多主观测试表明，观众对网络视频的感知质量不能仅仅由比特率和丢包率等常见的网络服务质量指标来决定。视频内容特征和压缩参数等因素也对感知质量有显著影响。本项目的目标是通过联合考虑影响感知视频质量的各种因素，在无线摄像机网络中实现高效的感知质量感知视频通信。这项研究的成功将提升用户体验，并通过基于无线摄像网络的应用程序增强用户探索和与物理世界交互的能力。本项目提出了一种无线摄像网络感知视频质量预测和控制的系统解决方案。提出的研究基于：i)对网络中感知视频质量进行轻量级且准确的预测；ii)综合分析感知视频质量与网络参数之间的关系；iii)探索利用感知质量的特性来实现能量效率和带宽效率的可能方法；以及iv)设计能够在动态网络条件下有效控制感知质量的新型码率控制和调度协议。该项目的研究方案有望在无线摄像网络中以最小的能量和带宽资源为用户提供令人满意的感知视频质量。研究解决方案将演示如何利用感知视频质量的特性来改善无线摄像头网络中的用户体验。这项研究将弥补感知视频质量研究与无线视频通信协议设计之间的差距，并将启发其他许多无线传感应用在感知质量保障方面的新研究。

Recent advances in imaging hardware and wireless communications have fostered the deployment of embedded camera sensors in various wireless imaging applications, such as surveillance, smart building operations, intelligent transportation, remote health care, and consumer electronics and entertainment. It is essential to guarantee the delivery of videos generated by camera sensors to an end user with good quality. In particular, the need to maintain good perceptual quality is driven by many human-centered imaging applications, where human users rely on the received videos to make critical decisions. However, many subjective tests have shown that the perceptual quality of networked videos as evaluated by viewers cannot be merely determined by common network QoS (quality-of-service) metrics such as bit rate and packet loss ratio. Factors such as video content characteristics and compression parameters also have significant impacts on perceptual quality. The objective of this project is to achieve efficient perceptual-quality-aware video communication in wireless camera networks by jointly considering the various factors contributing to perceptual video quality. Success of the proposed research will boost user experience and enhance user's capabilities to explore and interact with the physical world through applications based on wireless camera networks.This project proposes a systematic solution for predicting and controlling perceptual video quality in wireless camera networks. The proposed research is based on: i) light-weight and accurate prediction of perceptual video quality in the network; ii) comprehensive analysis of the relationships between perceptual video quality and network parameters; iii) exploring possible ways to achieve energy-efficiency and bandwidth-efficiency by utilizing the properties of perceptual quality; and iv) designing novel rate control and scheduling protocols that can effectively control perceptual quality in dynamic network conditions. The research solutions of the project are expected to provide satisfactory perceptual video quality to users using the minimum energy and bandwidth resources in wireless camera networks. The research solutions will demonstrate how to improve user experience in wireless camera networks by utilizing the properties of perceptual video quality. The proposed research will bridge the gap between the studies in perceptual video quality and the design of wireless video communication protocols, and it will inspire new research in perceptual quality provisioning for many other wireless sensing applications.