Research On Entity Data Caching Strategy Towards Internet Of Things Search

With the growing maturity of the global Internet of Things(Io T),cyberspace is further integrated with the physical world,and users,machines and devices are closely linked together.However,the explosive growth of widely deployed sensing devices in Io T has produced massive amounts of physical data.Faced with the massive physical data in Io T and the personalized and diversified requirements from users,search technology can be used to quickly and accurately search the physical data resources that meet the requirements.At the same time,researches have shown that caching physical entity data could realize the efficient search for entities.However,due to the massive,dynamic,and heterogeneous characteristics of entities,traditional Internet caching technologies have been unable to meet the real-time requirements of Io T search,and existing related researches have not focused on user search feedback and entity state changes.Therefore,efficient search matching cache strategy suitable for Io T has not been proposed.In view of the problems in the existing researches,this thesis has carried out related research on the entity data caching strategy towards Io T search.Firstly,the research background and significance of Io T search are introduced,and the research status of entities data caching in Io T search is summarized.Then this thesis briefly describes the concept,characteristics and application scenarios of Io T search,and analyzes the research situation of existing entity data caching strategies.Secondly,in order to reduce the frequent communication between users and data sources,the energy consumption of Io T perception devices and the delay caused by searching entity data,a reinforcement learning-based entity data caching strategy in Io T search is proposed.A responsive caching system framework adapt to Io T search is designed,in which reinforcement learning is used to solve the objective function and obtain the optimal caching decision so as to cache entity data that meets user requirements in the edge network close to users according to users' search feedback.Simulation results show that the proposed caching strategy is better than other baseline strategies,improves the cache hit rate and reduces users' search cost.Thirdly,considering the diverse time-varying features of physical entities,an edge-cloud collaborative entity state information caching strategy towards Io T search is proposed.Specifically,an entity state feature extraction method is presented to mine underlying rules via raw entity state observation sequence.Then,an edge and cloud collaborative entity state information caching strategy is devised to improve the search accuracy of Io T search system and reduce the search delay and energy consumption,in which entities are classified first according to the time-varying degree of their state and then these state information are discriminately cached based on their belonging classifications.Finally,the main work of this thesis is summarized,and the future research work is prospected.