Research On Distributed Intelligent Service Discovery Method Of Social Internet Of Things

With the promotion of 5G technology and the breakthrough of key technologies of the Internet of Things,various types of intelligent devices can be connected to the Internet of Things and provide corresponding application services for service demanders.At the same time,the progress of sensing technology and semantic technology makes data collection and annotation more convenient.The maturity of communication technology and network virtualization technology ensures the effective transmission of data information.The convergence and integration of data analytics and artificial intelligence has accelerated the arrival of the era of the Internet of Everything.With the rapid growth of service resources in the network,how to help users find personalized service resources quickly and efficiently,reduce service search space,so as to provide effective and accurate service discovery results has become an important problem to be solved urgently.This thesis researches on how to meet users' personalized,diversified and dynamically changing service discovery needs.The main work contents are as follows:First,this thesis proposes a time-aware personalized service discovery method under the social Internet of Things paradigm.In view of the problem that the service clustering method based on the similarity of service functions cannot describe the social relationship and dynamic changes of smart objects,the social similarity of objects is introduced to quantify the strength of the relationship between objects,and the similarity of object services is comprehensively considered.Perform service clustering with social similarity,reduce the scope of service search,and improve the efficiency of service discovery.Aiming at the problem that user service selection preferences are difficult to measure and the characteristics of preferences change over time,a user preference extraction model based on object usage behavior is proposed,and the influence of user preferences on the use frequency and average use time of smart objects is analyzed.Based on clustering,the service recommendation method is used to provide users with the results of service discovery that they are interested in.The simulation results show that compared with other service discovery methods,the method in this thesis has a greater improvement in accuracy,recall,and F1 value under different equipment scales.Second,in order to consider the user's preference for service Qo S and the objectivity of service quality to select the optimal service for the user,this thesis proposes an optimal service selection method based on service Qo S driving.First,the service Qo S is transformed into a unified standard in the same evaluation system according to different rules.Secondly,the user's preference for service Qo S selection and the objectivity of service quality are quantitatively expressed,and the proportion of the two in service selection is weighed based on the game theory.Finally,in order to obtain the best service among the candidate services,a Qo S evaluation model is established to evaluate the comprehensive capabilities of the service,and to compare each service's ranking.Experimental results show that the proposed method can effectively select the optimal service for users considering their Qo S preferences and actual quality of service.