无线网络中多视点视频的缓存放置问题

Content placement problem for multi-view videos in wireless network

在高速增长的无线多媒体业务需求量中,由于VR等相关设备的陆续出现使得在用户请求文件中多视点视频占据了越来越多的比重。与传统的二维视频不同,多视点视频有着较大的数据量,因此,将能够有效解决网络负载压力过大问题的缓存技术与多视点视频结合起来研究尤为重要。通过利用多视点视频基于深度图像的渲染技术（DIBR）将其与缓存充分结合起来。提出了一种缓存策略,在缓存容量有限的约束下,致力寻找出一种多视点视频的放置策略从而最大化缓存增益减小传输时间。通过一种有效的算法来解决文件的放置问题,在保证合成视点质量达到一定标准的前提下,将平均缓存增益最优化。最后通过仿真来证明该方法在解决多视点视频放置问题上的良好特性。

Multi-view videos occupy more and more proportions in requested files due to the successive appearance of related devices such as VR in the rapidly growing demand for multimedia services. Different from traditional videos, multi-view videos need more space. Therefore, it is particularly important to combine caching techniques that can effectively solve the problem of excessive network load pressure and multi-view videos. In this paper, we combine the caching with depth image based rendering technology（DIBR）in multi-view videos to reduce the storage space. The paper proposes an effective caching strategy to maximize cache gain and reduce transmission time with limited cache capacity. The placement strategy to maximize cache gain reduces transmission time. We have derived an effective algorithm to solve the file placement problem, and the paper optimized the average cache gain under the premise of ensuring that the quality of the synthesized viewpoint reaches a certain standard. Finally, the simulation shows the good characteristics of the method presented in this paper to solve the multi-view video placement problem.