Research On Service Architectures And Key Technologies For Smart Homes Based On Semantic Association Models

With the development of Internet of Things(IoT)and Artificial Intelli-gence(AI)technologies,various devices,sensors,Radio Frequency Identifica-tion(RFID)readers of Smart Home(SH)can form a new SH network.The SH network can be managed by an intelligent management system(e.g.,smart gateway and cloud gateway)and provide intelligent services for people.Tradi-tional SH technology usually customize corresponding SH solutions according to the individual requirements of different users.These SH solutions can en-able users to manage various smart devices deployed in the user's living envi-ronment through wireless communication or wired communication,effectively simplifying the user's life.However,this kind of SH technology depends on user's direct control and it is not smart enough to provide services for users.Therefore,a more intelligent SH system that can accurately recognize users'behaviors and automatically respond to the corresponding activities to satisfy users' actual demands is very desirable in the future SH market.However,the involved user behavior models in the existing approaches are usually con-structed with independent statistical probability of behavioral data.These mod-els cannot recognize the user's dynamical actions accurately and not suitable for providing assistance services to the elderly and the people with mild cognitive impairment.Moreover,the existing approaches usually do not consider the unavailable services,which causes the failure service recommendation.There-fore,in this dissertation,the research is focused on the new SH architecture,Smart City(SC)architecture and the key technologies which implement the device semantic models,semantic association(SA)model,relevant intelligent service algorithms and user privacy protection mechanism.The main contribu-tions of this dissertation can be summarized below:(1)An activity assistance approach based on SA model in SH.Each device in SH and its functions can be described by semantic models,hiding the heterogeneity,therefore these devices can understand and intercon-nect with each other.Different user has a different living environment,in which the configurations are associated with the user's requirements.The SA model is proposed through the semantic annotation analysis of semantic models.The SA model can make all the objects in SH establish associations in the logic layer,and provide services to users together.Based on the SA model,an ac-tivity assistance approach is proposed,in which smart gateway can understand and analyze the user's requirements when the related activities are detected and then provides appropriate services to assist people according to the current en-vironment status.The approach has been implemented and the results show that the success rate of the approach based on SA model is obviously higher than the approach applying predefined models.The proposed approach can effectively help people who are in trouble in the common life.(2)An Activity Recognition-Assistance(ARA)algorithm based on Hybrid Semantic Model in SH.While helping a user to understand and enjoy intelligent services,the smart gateway also need to consider the user's own living habits,which can be de-scribed by the statistical probability model.For the problem that the statisti-cal probability model cannot recognize the user's dynamical intentions accu-rately and misjudge the user's intention as a result,this dissertation proposes the hybrid semantic model based on the statistical probability model and the SA model,and an ARA algorithm is presented.The proposed algorithm is imple-mented in our living assistance prototype system.The algorithm first constructs a semantic association model for service recommendation,and then the user liv-ing habits are considered in statistical probability of Hidden Markov Model.In order to satisfy the user requirements of living habits,the algorithm will pro-vide the second service recommendation from the optional services,achieving a higher accuracy and better user experience.The simulation results show that,in the aspect of service provision,the ARA algorithm based on the hybrid se-mantic model is more accurate than the statistical probability model,and can effectively judge the user's intention and provide the corresponding service.(3)An Alternative-Service Recommending(ASR)algorithm based on se-mantic similarity.The dynamic on-line and off-line of SH devices will affect the smart gate-way to provide real-time services for users.However,the service recommend-ing models in the existing approaches usually do not consider the off-line status of unavailable services in users' living environments,resulting in the recom-mended service execution failed.To address the problem of dynamic service library,an ASR algorithm is proposed.The algorithm update the device status in the current environment in real time,and compute the semantic similarity ac-cording to the service library,which is abstracted from the on-line device.The algorithm can find the similar devices or services in the user living environment that can satisfy the same user requirements in order to provide an alternative service to maintain the normal living habits when the objective service is un-available.The simulation results show that,compared with the general service algorithm based on statistical probability models,the ASR algorithm can effec-tively find similar devices and services in the users' environments when some devices and services are offline,and satisfy the users' requirements.(4)Artificial Intelligence based Semantic IoT(AI-SIoT)service architec-ture and user privacy protection mechanism in SC.SC is the extension of SH.In the future,trillions of smart devices will join the SC and these devices usually come from different manufacturers with their respective protocols,so it will bring great challenges to the SC services.How to design an efficient IoT service architecture to integrate heterogeneous devices and provide intelligent services is the most important challenge in SC.This dissertation proposes AI-SIoT hybrid service architecture in SC,which involves AI technology and semantic technology.The AI technologies can im-plement effective data analysis and make appropriate decisions in intelligent services,and the semantic technologies can construct available service models for daily living assistance.Finally,this dissertation researches on the user pri-vacy protection in SC and proposes a mobile user privacy protection mechanism based on the shared data in the Cloud gateway.Users can find the required data in the cloud,and obtain the corresponding data access rights through match-ing their own identity attributes to decrypt data.The simulation results show that the attribute-based encryption algorithm based on the security classifica-tion access tree is more efficient than the traditional attribute-based encryption algorithm,and solves the efficiency problem of encryption and decryption with multi-attributes.