Research On The Strategy Of Virtual Resource Slicing Based On Characteristic Perception

With the consistent increasing of mobile users and the rapid development of communication and network technologies,the future wireless network is expected to cover all walks of life and form a large-scale interconnected and multi-service network.Therefore,limited network resources and heterogeneous service requirements have become the main challenges in mobile communication.To solve these problem,wireless network virtualization(WNV)technology realizes the dynamic sharing of infrastructure and wireless resources by multiple independent virtualized networks through network abstraction and segmentation,thereby improves the resource utilization and reduces the difficulty of network management.However,the virtualization of wireless resource and specific hardware which introduced by WNV is facing several problems such as wireless media uncertainty,user mobility,and heterogeneous wireless access technologies.For the implementation of slicing,the management,allocation,and sharing approaches of resources still require further research.Firstly,this thesis introduces the research background and typical application scenarios of virtualized wireless network,then briefly analyzes the current research hot spots.Secondly.Furthermore,the virtual resource slicing strategy in virtualized wireless network is researched,and the characteristics of existing resource slicing strategies are deeply analyzed and summarized.Secondly,a virtual resource slicing strategy based on user clustering is proposed.By perceiving the network,service and social characteristics of network slices,the gain and expenditure of different slices are mapped into a unified model,thereby developing an optimization framework for resource slicing.Then,according to the subjective preference and resource requirement of users,they are divided into different characteristic groups to perform service differentiation resource allocation.Furthermore,the social relationship between users is perceived,then the cellular automaton model is used to form user clusters and dynamically update and group status.Finally,a heuristic resource allocation algorithm based on user clusters is proposed to realize efficient virtual resource slicing.Thirdly,a virtual resource slicing strategy based on user context awareness is proposed,which establish a heterogeneous resource model between network slices,and a resource slicing framework based on perception-decision-feedback is formulated based on user characteristics.Then,the personalized service situation of users is established by evaluating the service performance parameters,and the Quality of Experience(QoE)metric is introduced,by perceiving the service context relationship of users through the Bayesian Networks(BNs),the QoE measurement model is established.Furthermore,the heterogeneous utility function is dynamically adjusted according to the real-time feedback from users,and the genetic algorithm is used to find the optimal decision through all the candidate resource allocation decisions,thus the intelligent slicing of virtual resource is achieved.Lastly,the research works are concluded and the follow-up works are prospected.