全景视频给人以浸入式和在场式视觉体验，从而被广泛认为是未来的视频格式。但是，其超高清的多视角视频格式，需要很大的网络带宽，对发送端设备、终端设备及数据网络造成很大的资源消耗。本课题提出预测式交互全景视频研究方案：设计全景视频可扩展式编码方案；解决信源信道编码的多维数学优化问题；创新性的对全新的全景终端用户行为进行数学建模和预测。其预先传输用户感兴趣的视频区域码流，使得保证接收端低延迟的前提下，最大化的降低分发全景视频数据的代价。课题预期为未来全景视频的无线分发提供技术基础。

Panoramic video enables immersive and presence visual experience, which is expected to be the future video format. However, extremely large bitstream is necessary to represent the multiple ultra-high definition (UHD) views of panoramic video, which exhausts the resources of the source host, the network and the terminals. In this project, we propose the so-called predicted interactive panoramic video research methodology, which investigates the solutions of a flexible panoramic video coding, multi-dimensional optimization problem for joint source-channel coding, modelling and prediction of the novel terminal user behavior. Specifically, we predict the region of interest (ROI) of the users and deliver the corresponding bitstreams in advance, thereby minimizing the consumed resources without imposing extra delay. This research will pave the road for the wireless delivery of the future panoramic video.