Research On Several Key Issues In Data Processing In The Internet Of Things

Every major breakthrough in information and communication technology (ICT) has brought a boost to the development of society. Since the birth of the first computer ENIAC in 1946, ICT undergoes a rapid development in nearly seventy years. Internet has deep impacts on all aspects of human life in the long run. The emergence of the internet of things (IoT) is widely considered as a significant advancement following the information technology revolution of Internet. The IoT has been launched broad demonstration applications in various fields and frequently investigated in research areas around the world.The IoT is a vital extension of Internet for enabling comprehensive connection to the physical world and building perception infrastructure in pervasive environment. Meanwhile the IoT arises information ecosystem among human world, physical world and information world which comprised network, information and service. Internet has facilitated communication and been sharing information between people. Further, through the continuous development of the IoT, it will enable information ecosystem between the human and physical world closely. Thus, it will change traditional utilization of Internet for information communication and resource sharing.This dissertation summarized characteristics of the data and information in the IoT from the perspective of evolution of the IoT. Our main focuses on key issues in data processing in the Internet of Things are based on characteristics including heterogeneity, massive amounts and real-time. The main contributions of this dissertation are detailed in the follwing.First, this dissertation aims to review the evolutionary process of the IoT. This dissertation conducted that from the perspective of correlative technologies and presents the process in a chronological order. Through generalizations of particular focus in different stages of each technology, researchers can better understand the current phase of the IoT, and therefore, predict future challenges. Information on evolving of the IoT is missing in the current literature. It focuses more on the introduction and comparison of existing technologies and less on the evolutionary process of the correlative technologies. The latter is fundamental work to select a proper research angle of the IoT.Second, in the view of data characteristics with regard to heterogeneity and massive amounts, it also requires interoperable service-oriented technologies to share real-world data among heterogeneous devices, and then integrate and fuse these multi-sources heterogeneous IoT data. In order to address these issues, architecture of information fusion in the IoT is necessary to be proposed, which can illustrate guidance for the development of information fusion in the IoT. In this dissertation firstly compared features regarding data and information in the IoT with existing wireless sensor network and Internet; and as far as we knew, the comparison is first raised by us. Then this dissertation designs a SOA framework for multi-sources heterogeneous information fusion in the internet of things based on semantics. Meanwhile in the framework of the SOA middleware this dissertation designed a type of semantic annotations service layer for the IoT heterogeneous data service. By levering semantic annotation operation, this dissertation proposes a semantic heterogeneous data fusion algorithm for realizing the interconnection and interaction between Internet and other resources. Then we use leverage an experimental simulation platform to build an environmental monitoring system for verifying.Third, the majority of data in the IoT is streaming data which produced by various types of sensor in real environment.These big streaming data application is with the requirements of high real-time performance and low latency. In order to address these issues of real-time, we design a real-time method to handle streaming data of IoT with low latency. This dissertation proposes an online approach to addresses the real-time issues, which focuses on employing only real-time data rather than mining historical data. Thus, the proposed method does not require extra time on the statistical analysis of historical data and avoiding the time consumption of data mining on historical data. Furthermore, we set up a real-time system to process streaming data of IoT with low latency. The results suggest that the system and the proposed approach can effectively operate and guarantee correctness, low latency, scalability, fault tolerance, state consistency, load balance and high throughput.