Study On Hierarchical Architecture Based On Rule Reasoning And Edge Computing In Smart Environment

In recent years,people yearn for a fast and comfortable living environment,so that the smart environment has sprung up under the support of technology for pervasive computing and the Internet of Things.Smart environment is an indoor environment with a variety of data acquisition devices,which combines various technologies to conduct reasoning and analysis of data,and then the conclusions are used to realize the functions of activity recognition,data perception and control.At present,the smart environment seems to have broad prospects for development,as it has the characteristics of strong industry penetration,wide range of users and rapid development of related technologies,however,it has the shortcomings hided of single product function which is difficult to meet user needs,weak portability and bad real-time performance.In view of the above problems,after analyzing the current situation of related technologies,this thesis determines to build a smart environment system by way of sensor recognition and driven by knowledge.This thesis mainly includes the following work:(1)Design a hierarchical architecture,which clearly divides each function module.Separate the "smart layer" to divide the acquisition,representation and perception reasoning of context data,therefore,the underlying implementation is shielded so that the rapid modeling and portability of the system are enhanced.(2)The production rule reasoning framework implemented by Rete algorithm is integrated into the above-mentioned hierarchical structure,and the rule compilation module is constructed to separate the code from the user-customized rules,which better meets the flexible configuration of user demand.(3)In order to further improve the real-time performance of the system,this thesis uses the idea of emerging edge computing to design the algorithm for allocation of inference nodes to migrate the reasoning work of the central gateway to the smart nodes,which combines the characteristics of rule reasoning using Rete algorithm in the smart environment.The proposed mechanism overcomes the system delay,in the meantime,edge computing also plays a role in isolation and protection of privacy data.(4)As for the rule compiling module and reasoning module in the hierarchical architecture,they are implemented in the BeagleBone development board to verify the accuracy of the results;As for the proposed allocation algorithm of inference nodes,the evaluation parameters corresponding to real-time performance and resource utilization are designed,which are observed after setting different conditions in the simulation environment,therefore,the validity of the allocation mechanism is verified.