Research And Design On Self-Configuring Smart Home System Based On Semantics

The rapid development of smart home appliances and intelligent home services has brought several benefits to its users.However,there are some major problems that arises in this domain.Aside the device interoperability difficulties caused by device heterogeneity,the interaction between applications and devices still relies on the pre-set solution and the users' manual intervention.Therefore,the existing smart home implementations cannot adapt to complex and dynamic home environments which renders it short of meeting critical user requirements.For these reasons,a semantic-based self-configuring smart home system is presented in this thesis.This smart home system is able to perform adaptive operations in the smart home environment according to the service requirements and user preferences on the basis of the semantic interconnection of devices to provide a comfortable living environment.First,we analyze the characteristics of smart home devices and model the devices from four perspectives: function,operation,state and characteristics.These perspectives are based on the information that the devices serve in the decision-making process.This model is a unified description of device information used in the decision-making process and it provides benefits for the interconnection of devices.Since the interaction method between devices and applications is single and fixed,and cannot changes dynamically based on the service demands,we propose a concept of a service entity which is regarded as the linker and mediator between an application and a device.Base on application requests,the service entity is able to schedule the relevant devices to implement environmental adjustment and control.We also propose a general semantic model of the service entity,which facilitates interoperability among service entities.Moreover,the service entity associates devices to environment attributes,and facilitates the linkage of devices and the observation-control loop.In order to facilitate the self-configuration of users,we built a unified user ontology model.In this thesis,we treat the service entity as an independent decision maker and executor,and use the semantic information and rules in the smart home system to instantiate the MDP decision model parameters,including state,action,transfer probability function,reward function and objective decision.Moreover,we propose the Feedback MDP algorithm(FMDP)which is able to adjust parameter values of the reward function and the state transfer function of the strategy adopted by the service entity based on the user feedback.A smart home scenario was used to design and validate the semantic-based selfconfiguring system.First,we propose the functional requirements and system architecture and the main functional modules are designed in detail including data entry,semantic process,rule management,and MDP-based dynamic decision-making process.The verification of the semantic-based self-configuring system takes the form a comparison between a normal home automation system and our proposed system.The results of this comparison demonstrates the feasibility of the proposed system.