| With the rapid development of Virtual Reality(VR)technology,new virtual reality terminal devices and services are increasing day by day.The comprehensive network applications supporting virtual reality services make network resources more and more scarce to meet users’ high-quality needs.Virtual reality may completely change the way users interact with media,breaking through the traditional multimedia only a single passive acceptance of the experience model.However,there is a relative lack of research on the quality assurance of our VR services,so research on the quality assurance mechanism of VR services is necessary.We need to better understand the impact of user attributes on the quality assurance of VR services and build a quality assurance mechanism for VR services based on the perception of user attributes,which is one of the important issues that need to be addressed urgently during the development of VR services.Existing studies have focused on the optimization methods of demand metrics during VR network applications,but when faced with complex and variable user demands,the optimization of a single metric usually has limitations.In the future development of VR services,virtual reality services place more emphasis on user participation and the way they interact with VR content,going beyond the passive mode of traditional video services,and the characteristic attributes of users will directly affect the service quality of VR.However,the current issues of VR service quality assurance based on user perspective have been little explored.Therefore,this paper will systematically study the problem of VR service quality assurance from the granularity of user characteristics attributes.Among them,this paper mainly considers that the mapping model between service quality of service(QoS)metrics and user quality of experience(QoE)brought about by the lack of multiple user attributes cannot well evaluate the user-centered VR service quality objectively.At the same time,the user AP association method and content caching policy in the VR user network application process will not be able to guarantee the impact of network latency and network throughput indicators on service quality due to the negligence of user behavior attributes and user VR content attributes.The lack of consideration of user social attributes also limits the quality of user experience when facing the scenario of multi-user service demand.In this paper,the quality assurance mechanism of VR services is studied in depth from four aspects,namely QoE-QoS mapping,AP association,AP caching,and multicast optimization,with the overall goal of optimizing user experience.The main research contents and results are summarized as follows.(1)In response to the lack of objective user experience evaluation methods and models in the current VR service assurance mechanism,a QoE-QoS mapping model with the goal of improving VR user experience is proposed.The model integrates the main parameters of subjective and objective factors of VR user viewing experience,builds a QoE evaluation framework based on the subjective feelings of VR users,and uses fuzzy hierarchical analysis to extract QoS parameter weights for this framework.At the same time,a QoE-QoS model is established based on psychological evaluation methods combined with parameter weights,and an optimization model incorporating deep learning methods for user personalized features is proposed.The experimental results show that the effectiveness and accuracy of this mechanism is better than the traditional QoE-QoS mapping model.(2)To address the problem that the current AP association method only focuses on the performance improvement of common users and ignores the impact of VR user behavior on service quality,an AP association method based on VR user behavior awareness is proposed.Under the constraints of AP multi-rate and load balancing,the optimization problem of minimizing the average download delay of users is constructed,and a heuristic algorithm is proposed to construct a local optimum to achieve the optimal AP association policy for VR users.Simulation experimental results show that the proposed method exhibits better results for both average user download latency and load balancing metrics compared to the baseline algorithm.(3)To address the problem that the current AP caching policy only focuses on the popularity of general video users and ignores the VR video features for service performance improvement,an AP caching policy based on VR user content awareness is proposed.The policy takes into account the global content popularity and local content popularity of user-requested VR content and the prominence of VR content itself,and constructs it as a combinatorial optimization problem,and then proposes a heuristic algorithm to find the approximate optimal solution of the problem.Finally,simulation experiments are conducted to verify the effectiveness of the proposed algorithm by analyzing the sum of network throughput and the cache hit rate under different cache spaces.(4)The multicast optimization algorithm based on multi-user social attribute aware VR video is proposed to address the problem that the current VR multicast technology lacks the consideration of user social attributes possessed by multiple users.The mechanism takes into account the social relationship of the social domain,the similarity of user behavior and user content preferences,builds a user multicast group with social attributes,designs a greedy user group multicast algorithm based on iterative ideas,and further combines group iteration to maximize the utility of the system as a whole on this basis.Simulation experiments show that the mechanism achieves substantial improvement in the overall system effectiveness and user experience evaluation compared with the classical random algorithm. |
