Research On Middleware System Of Internet Of Things Based On OCF

The rapid development of Internet of Things(IoT)technology promotes its application on a great range of areas.The main challenge on application of IoT is the communication between multi-scene IoT services.Therefore,middleware technology was developed.This thesis focuses on diversified IoT devices’ access and service management,and designs OCF-based IoT service middleware system,and also researches on access identification of heterogeneous devices in middleware system and key technology of serialized service call method,providing support on achieving service-sharing technology for upper-layer IoT applications.The main works of this thesis is as follows.First of all,for the problem of bidirectional communication normalization,this thesis researches the OCF protocol.Combining the principal of service-based structure,this thesis designs the middleware system for IoT services.The thesis also proposes a middleware system with device access and service managing ability.We,and plugs heterogeneous devices into the middleware system to identify,abstract them and project them as concrete services.By making use of the middleware system’s unified service description and service call,this thesis achieves diverse services cooperation of upper-layer applications.This thesis provides procedure norm for devices access and services management of middleware system.Second of all,for the problem of accurate identification of IoT devices in the middleware plugging in procedure,his thesis proposes device access identification method based on passive finger print.On the device configuration stage,this thesis collects and analyze IoT devices’ network data stream,extract data stream’s protocol feature and statistical feature,and import SMOTE algorithm to solve the data sample imbalance problem.Based on these,with random forest algorithm,this thesis constructs passive finger print to accurately identify and extract device type,service class and so on.A large amount of experiments prove that this method can improve device identification accuracy and effectively support the middleware system’s device plugging in ability.Last of all,for the problem of storing and transporting a great number of data when calling middleware services,this thesis designs service data optimization and serialization solution.On the devices serving stage,by cooperation of middlewares,this thesis achieves management and mutual calling of plugged in devices and their diverse services,besides,this thesis model the service calling procedure to form the serialization problem targeting on joint minimization of latency and power consumption.This thesis decomposes this problem into two subproblems to solve.This thesis propose machine learning based serialization method selection algorithm and Lagrange multiplier based serialized executing decision algorithm,to extract device service data pack feature to select optimal data serialization solution for data packs with different features.The following experiments prove that our method can reduce latency and power consumption,and therefore effectively improve the middleware system’s service calling efficiency.