With IT and video technologies evolving rapidly, video services with user interaction emerge, and bring about attracting experiences. The growing bandwidth and latency demands call for advanced transmission techniques which utilize wired, wireless, and 5G cellular network, and cooperate unicast, multicast, and broadcast. Free-viewpoint video, as one of the interactive video applications, can provide users with immersion of real world scenes, and shows potential to be an essential application in future digital life. However, such an application is different from other video services in aspects of computation and communication features, and existing researches are not up to the low frame latency, smooth playback, and high service scalability challenges. In this paper, we target on the above problems, and achieve ultra-low latency and stable video transmission for free-viewpoint video services by improving the performance of multiple system components accordingly. Firstly, the key performance indicator is formulated for such services, and an edge computing based video transmission schema is proposed to achieve ultra-low frame latency. Secondly, a pair of stabilization algorithms at both server and client sides based on application features are introduced to ensure stable encoded video streams before transmission, and smooth playback after that. Thirdly, a cloud-edge-client hybrid free-viewpoint service architecture is proposed, which can provide hierarchical experiences and enlarge the service scale with restricted resources. Experimental results prove the effectiveness of the proposed transmission schema and scheduling algorithms, and feasibility of the proposed service architecture is revealed by qualitative analyses.

中文翻译：

---

超低时延、稳定、可扩展的视频传输，满足自由视点视频服务


随着IT和视频技术的快速发展，具有用户交互功能的视频业务应运而生，并带来了吸引人的体验。不断增长的带宽和延迟需求需要先进的传输技术，利用有线、无线和5G蜂窝网络，并配合单播、组播和广播。自由视点视频作为交互式视频应用之一，可以为用户提供真实世界场景的沉浸感，有望成为未来数字生活的重要应用。然而，此类应用在计算和通信特性方面与其他视频服务不同，现有研究无法应对低帧延迟、流畅播放和高服务可扩展性的挑战。本文针对上述问题，通过相应提高多个系统组件的性能，实现自由视点视频业务的超低延迟和稳定的视频传输。首先，制定此类服务的关键性能指标，并提出基于边缘计算的视频传输方案，以实现超低帧延迟。其次，根据应用特点，在服务器端和客户端引入了一对稳定算法，以确保编码视频流在传输前稳定，传输后播放流畅。第三，提出了云边客户端混合自由视点服务架构，能够提供分层体验，在资源有限的情况下扩大服务规模。实验结果证明了所提出的传输模式和调度算法的有效性，并通过定性分析揭示了所提出的服务架构的可行性。