Development And Performance Optimization Of 5G Wireless Virtual Reality Application Test Platform Based On Fog Computing

Virtual reality(VR)provides users with immersive viewing and interactive experience by generating 360 degree scenes,involving stereo display,rendering,interactive perception and other technologies.However,the traditional local VR scheme is restricted by the high hardware cost and user mobile experience,while the emerging cloud wireless VR scheme faces the challenges of large-scale data transmission and long delay.Therefore,based on the 5th generation mobile communication technology(5G)and the concept of fog computing,this paper designs and implements 5G wireless virtual reality application test platform based on fog computing.A series of performance indexes are tested and analyzed for wireless VR applications,and a joint optimization method of application layer and wireless network layer parameters based on off-line deep reinforcement learning is proposed.The main work and contributions of this paper are as follows:1.Aiming at the problem that the VR scheme based on special highperformance host and wired connection and the traditional cloud wireless VR scheme are difficult to achieve both lightweight,mobility and highquality VR scene,a 5G wireless VR application test platform based on fog computing is designed and implemented.The platform includes 5G transmission module,fog computing module,VR application module and platform management module.The core feature is to unload the graphics rendering computing of VR scene to the fog computing module,and then use 5G wide bandwidth to stream the rendered encoded VR scene frame to the user terminal.Because the terminal only needs to carry out simple secondary rendering and scene playback,it not only reduces the computing pressure and ensures low delay and high-quality transmission,but also enables users to get rid of the constraints of wired and significantly improve the user experience.In addition,in order to monitor the typical performance indicators of VR and optimize the VR user experience in the fluctuating network environment,the interactive interface between the platform management module and 5G transmission module and application layer module is developed to realize the collection of network and application status data,which can effectively support the training of artificial intelligence model and the distribution of parameter configuration for user experience optimization.2.Aiming at the problems that the performance evaluation index of wireless VR application is relatively single and the optimization scheme fails to fully consider the impact of VR application layer and wireless network layer,this paper analyzes the application characteristics of strong interactive VR application by using the built platform,including the performance requirements of wireless network bandwidth and delay.In addition,through the evaluation of application performance indicators under cloud and fog computing,the supporting advantages of fog computing on wireless VR application performance are displayed.At the same time,the effects of VR application layer parameters and wireless network layer parameters on application performance are tested and analyzed.Based on the above tests,aiming at maximizing the quality of VR user experience and reducing the use of base station radio resource blocks as much as possible,this paper proposes a multi-layer parameter joint optimization method based on off-line deep reinforcement learning.In the scenario of two VR users,the effectiveness of the proposed method is verified by comparing with the traditional Q-Learning and random parameter selection strategies.In addition,by comparing the multi-layer parameter joint optimization scheme with the scheme that only optimizes the parameters of application layer or wireless network layer,the performance advantages of multi-layer parameter joint optimization are intuitively displayed.The results show that after the multi-layer parameters are jointly optimized,the use of radio resource blocks in the base station is reduced by nearly 30%,and the parameters such as VR application layer resolution and bitrate are also improved to varying degrees.To sum up,this paper designs and implements a VR demonstration platform based on 5G transmission and fog computing,tests and optimizes the performance of strongly interactive VR applications,and puts forward a wireless VR application performance optimization method based on offline deep reinforcement learning,which improves the quality of VR user experience and significantly reduces the occupation of base station air interface resources.