Software Defined Network And Service

Software defined network(SDN)is a novel architecture that redefines the network structure in a subversive way.SDN is expected to meet the increasing needs of complex network data interaction,which is a research hotspot.SDN defines the northbound and southbound interfaces to decouple the control and data planes,and realize flexible control and management of network for optimizing network performance.This thesis focuses on the SDN-enabled applications in the field of healthcare and Internet of things(IoT)and etc,specifically including software defined healthcare network,software defined IoT and semi-physical simulation,software defined personalized recommendation and software defined caching strategy in Named Data Networking(NDN).In the aspect of software defined healthcare network,the existing problems and limitations of medical services based on IoT are comprehensively investigated,and a software defined healthcare network(SDHN)architecture is proposed,which consists of physical layer,control layer and application layer for realizing the uniform persistent data management and decreasing the complexity of service access.Moreover,a testbed for a typical healthcare application is developed to evaluated the quality of service of SDHN.In the aspect of software defined IoT and semi-physical simulation,SDN is involved to redefine the WSN framework considering the limitations in the exsiting solution of IoT.In this paper,the design of northbound and southbound interface of SD-WSN architecture are presented,and a semi-physical simulation integrating real WSN network and simulation network is described.Furthermore,a simulation based on healthcare is designed to verify the core indicators of SD-WSN framework.In the aspect of software defined large e-commerce recommendation system,a novel software defined personalized recommendation(SD-PR)architecture is proposed.Based on SD-PR,an improved collaborative filtering algorithm using smoothing and fusion(CFSF)is developed to optimize the precision and real-time of recommendation system.Compared with the related approaches,the experimental results show that CFSF has better performance in scalability,real-time and accuracy.In the aspect of software defined caching strategy in NDN,an overview of the defects of content-oriented caching strategies based on existing architecture in NDN is provided.From this point of view,combined with the flexibility of SDN in network management,a novel software defined caching mechanism is proposed to fully consider the number of nodes and the properties of communities.Moreover,a prototype system is developed to verify the proposed model,and the simulation results show that the proposed software defined caching strategy has better network performance.