Reinforcement learning-based edge prefetching method and system for vr video in c-ran architecture

Abstract

The invention discloses a VR video edge prefetching method and system based on reinforcement learning in a C-RAN architecture. The method includes: collecting network throughput, user request information and cache status information in real time; based on video quality, video time domain jitter, Video spatial jitter and delay determine the user quality of experience of a single user, and predict the user quality of experience gain of a single user; determine the user quality of experience gain of multiple users based on the user quality of experience gain of a single user; Optimize the quality of experience gain; edge prefetch for VR video based on network throughput, user request information, cache status information and optimized multi-user user quality of experience gain. The present invention can adopt the method of dynamically prefetching multi-level cache in C-RAN to reduce time delay and reduce repeated data transmission, thereby providing users with a more comfortable VR video viewing experience.

Description

technical field [0001] The present invention relates to the technical field of computer networks, in particular to a VR (Virtual Reality, virtual reality) video edge prefetching method and system based on reinforcement learning in a C-RAN (Cloud Radio Access Network, cloud radio access network) architecture. Background technique [0002] With the development of VR technology, VR has gradually entered thousands of households, providing users with immersive video viewing experience, and also playing an indispensable role in the fields of educational interaction, industrial remote guidance, and telemedicine. According to statistics, as of 2019, the number of VR users in China has exceeded 10 million, and the industry revenue of virtual reality software and hardware has exceeded 1 billion yuan. It is foreseeable that virtual reality technology will flourish in the future and will expand to more In the field of application, it can provide users with a more realistic, high-definit...

AI Technical Summary

Problems solved by technology

For example, transmitting 8K panoramic VR video requires a bandwidth of more than 260Mbps, and if you want to obtain a more extreme experience effect, you need a bandwidth of more than 10Gbps, which is a huge challenge for the current network, especially the backbone network of the video source server. challenge

On the other hand, users are very sensitive to the delay of VR videos. Generally speaking, if the delay exceeds 20ms, they will have a strong sense of dizziness, which is disastrous for the viewing experience of VR videos.

Images