S&T | The Concept of Data Fusion in the Internet of Things

S&T | The Concept of Data Fusion in the Internet of Things

By Xiaotie Qina and Yuesheng Gub

S&T | The Concept of Data Fusion in the Internet of Things


The application prospect in the market with huge thing networking are buzzing the third wave of information technology, its one of the core technology on two wireless sensor networks with energy, storage capacity, computing power, communications bandwidth resource constraints of the salient characteristics of data fusion, implementation is the inevitable choice. This paper briefly describes the Internet of things in the characteristics of data acquisition and data fusion of things in the meaning and role of data fusion described things to be addressed key issues and requirements. Briefly describes the data fusion tree formation and fusion of the principles of TGA

Main text

Internet of things is connected with all kinds of wireless access network through the interface, and access the internet connected to objects, can achieve intelligent endowed with object communication and dialogue, also can realize the interaction between objects and object communication and dialogue. To the object have overall perception. To transfer data are reliable and intelligent processing ability of the information network by connecting object and the object[1].

In a Internet of Things, to perception nodes, nodes and gathering nodes and Between the control center, they not only need to communicate but also need do further treatment with the communication results. Therefore, data fusion is one of the most important technologies to internet of things. In internet of things, Using Radio frequency identification device, the global positioning system, laser scanner and all kinds of different devices, embedded software and hardware systems, and the modern network and wireless communication, distributed data processing, and many other technology. Collaboration is real-time monitoring, perception, collecting network distribution area of various environmental or monitoring object information. To realize things and material, including between people and objects of all kinds of interconnect. Combined with the Internet and formed a huge information network system. In the front of the networking of things, for example sensor network, in order to obtain accurate data, often need to deploy a lot of sensor nodes in monitoring area internal. In the sensor node monitoring range fold each other in order to enhance the whole network the accuracy of the information collected. Covered in high density areas, for the same object or event monitoring of adjacent nodes getting the report data, there will be a certain space correlation. As distance similar node the transmission has certain redundancy. If all nodes will be monitoring data sent to the focal point, it can cause the network bandwidth limited tremendous waste of resources. And large amounts of data transmission at the same time can also cause frequent conflict, reduce communication efficiency. In addition, data transmission when the main factors of node energy consumption, transmission of redundancy data will consume too much energy envoys from point, so as to shorten the life cycle of sensor network. Therefore, in a large sensor network, each node more data to jump transmission perception together before, need to node data fusion processing.

Author Artwork

In the process of information perception, make full use of local node of the computing power and storage capacity, for a lot of data or information processing, the combined from more effective and more in line with the needs of the user data process method. The study of data fusion technology in internet of things, in addition to the basic content of data fusion outside, need to focus on the choice of the points to fusion, fusion timing of data fusion and how to do data fusion. There is a close relationship between the choice of fusion node and the network layer routing protocol. Need to rely on a loop routing protocol data, and the use of some nodes routing structure as a data fusion node. Internet of things is just like the sensor network. It is a kind of jumped the organization network, perception area-based data aggregation node need cooperation. Especially in periodic monitoring applications, the need to consider it. Periodic perception node data. Adjacent rounds of data acquisition has some relevance, need information such as historical information to reduce the amount of data it. When determined the path of data aggregation and data fusion points after data fusion energy saving effect also and these data fusion nodes data fusion of waiting time before closely related. Need to know for how long, and what the data from node merge. Namely needs to determine the appropriate data fusion node data fusion opportunity.In a data fusion node, when and to what the received data fusion and forwarding, requires a combination of routing protocol to forward mechanism to consider. In data aggregation, need routing as far as possible to transfer some of the packet network nodes, and data fusion in these nodes. What kind of routing algorithm will directly affect the efficiency of data fusion. Data fusion is in order to adapt to a sensor network center for application in data from the fusion algorithm, the main focus on how to use local perception of processing power nodes, and then the data processing it, the focus is to reduce the data needs to be transmitted.

It needs to improve the efficiency of information from the perception in internet of things, and it will be based on data fusion triggered the connection. Considering the perception of the node will due to bad environmental factors or their energy exhausted and cause failure, so the case such as robustness and adaptability is to realize basic requirements for data fusion in internet of things.

In internet of things, because of perception of communication between the nodes may cause interference, and networking information perception may be no accuracy, so pay more attention to solving the related data ambiguity. In internet of things, network nodes energy Co., LTD, del and send and receive data node energy consumption far outweigh the computing and storage energy consumption. We need to consider the perception of the energy consumption and energy node balance, how to choose the right to study and solve the fusion processing nodes. Only to solve the technical index, we can acquire the correct and effective data.

Data fusion algorithm is the basic content of the fusion processing, it is many-dimensions nput data fusion in different level with different mathematical method, and it is the clustering processing method. At present, according to data fusion in the sensor network , the researches focus on the data fusion routing protocol both at home and abroad. Based on data fusion tree is to compare the routing protocols of the typical data fusion routing algorithms, it is the key to how to construct data fusion tree.

When a sensor nets, base station or collect data gathering node is through reverse multicast tree from the perception of the form of scattered data gathering up gradually nodes. When each perception node to sudden emergencies, monitoring data transmission path form a tree reverse multicast tree, the number is data fusion tree. We can structure data fusion center tree through the shortest path tree or recently source node algorithm. The shortest path tree from the base station or focal point to recent source node ACTS as the fusion center node. All the other perception node first put the data collected the minimal hop along the path to the node, then sent by the node will be merged data base station or gathering of hair to node. Once you have identified the fusion center node tree is determined, the integration of basic down. CNS algorithm is actually a fusion of priority data fusion tree structure algorithm, in the specific implementation we need to know which nodes from gathering node recently, and other source node to the node routing information. In the middle of the upper nodes are data fusion. When all source nodes of the shortest path are determined,fusion tree has the basic form. SPT data fusion algorithm and the number of points are random, it is actually a routing priority trees make a data fusion algorithm, and is easy to realize in the specific implementation, distributed can draw lessons from Ad Hoc network when routing protocol thought.

TGA (of aggregation) is a kind of data-centric fusion tree based on data fusion routing algorithms, mainly by inquires the distribution and data receiving of two stages. Inquires the distributing stages: gathering node send inquires the perception node, requested and complete data fusion tree structure, the gathering node to adjacent nodes through radio data fusion tree structure news to construct the assemble the node is the root of the trees, the tectonic news data fusion contained in the hierarchical information in tree node, have not joined the node to data fusion tree after receipt, tectonic information for constructing the level set the level value plus 1 news, and will construct news for father sent node set node. In data collection phase, along the collected data nodes father node direction sent to data fusion the roots of a tree node. Node to adjacent node radio while the tectonic news; Data receiving stages: each node of the data collected after fusion of the structure by sending gathering data fusion tree data fusion tree node, and in each of the father node must gather all the child nodes of data and fusion to data fusion trees along the fusion data sent to converge node. Meta can use database query language, complete the COUNT, MIN, MAX, from the AVERAGE wait for basic, data fusion operation. Meta's weakness is vulnerable to network topology structure dynamic changes and the influence of link failures should periodically, gathered node tree structure news radio data fusion to maintain data fusion trees, to adapt to changes in the network topology institutions. When the network topology structure, such news frequent change of transmission will consume too much energy, shorten the life cycle of sensor network.

Directional Diffusion (DD) : Diffusion-ship is a kind of typical data-centric based on data fusion algorithm fusion tree. The directional diffusion routing fusion algorithm data fusion including interest spread, interest gradient establish and data sent through three stages.

When gathering to collect data, nodes at first to broadcast the sensor node net Interest news (macro), interesting Message describes gathering node interested data types and data acquisition methods in receiving the information such as the Interest, the nodes, but also the news to the nearby nodes radio interesting Message. Node according to accept degrees aode interesting message establish interest gradient, contains the node used in interest gradient the next jump forward packets of information, the interest gradient node is according to the message sending node energy, communication ability, position adjustment factors. When interest diffusion process, each node will end according to interest gradient value choosing an "enhanced path" as data transmission path. When the node with interest to collect data, matching the data sent to enhance path next jump node. Accept nodes via caching mechanism to reduce data redundancy, when data accept node receiving data and preservation found in cached repetitive already forwarding data will not be forwarding, to reduce redundant data delivery.

Directional diffusion algorithm by functional features without the node way to getting the whole network flood topology information, only through the interaction with neighbors node can build and interest gradient and used for data forwarding enhancement path, improve the network resource utilization. But directional diffusion algorithm also need periodic updates data fusion tree, to adapt to the dynamic change of network topology.


In short, through data fusion in internet of things, it can reduce things networking traffic, reduce the energy consumption and enhance the accuracy of the results, optimize the perception that the network performance.  

About The Authors:

Xiaotie Qina - Library, Chongqing University of Science & Technology, Chongqing, 401331, China
Yuesheng Gub - Henan Institute of Science & Technology, Xinxiang, 453003China

Publication Details:

1877-7058 © 2011 Published by Elsevier Ltd, Open access under CC BY-NC-ND license, doi:10.1016/j.proeng.2011.08.567 / Download The Paper - LINK


[1]LiuHuaJun. Internet of things architecture research [J].journal of Chinese new communication, 2010 (5) „

[2]lee. Wireless sensor network theory, technology and implementation [M].beijing: national defense industry press, 2008 „

[3]YangWanHai. Multisensor data fusion theory and its application [M]. Southwest: xian university of electronic science and technology press, 2004 „

 [4][U.S.] Klein, Lawrence A. The multisensor data fusion theory and application [M]. DaiYaPing, etc. Beijing: Peking up, translation polytechnic university press, 2004 „

[5]ChenLinXing. Wireless sensor network technology and application [M].beijing: electronic industry press, 2009 „

[6]Hongbo, HuangStudying on internet of things based on fingerprint identification . ICCASM 2010 - 2010 International Conference on Computer Application and System Modeling, Proceedings, v 14, p V14628-V14630, 2010 „

[7]Huang, Yinghui .Descriptive models for Internet of things. Proceedings of 2010 International Conference on Intelligent Control and Information Processing, ICICIP 2010, n PART 2, p 483-486, 2010, „

[8]Song, Zhexuan.Semantic middleware for the internet of things. 2010 Internet of Things, IoT 2010, 2010, 2010 Internet of Things, IoT 2010 „

[9]Wu, Miao.Research on the architecture of Internet of Things. ICACTE 2010 - 2010 3rd International Conference on Advanced Computer Theory and Engineering, Proceedings, v 5, p V5484-V5487, 2010 „

[10]Atzori, Luigi.The Internet of Things: A survey. Computer Networks, v 54, n 15, p 2787-2805, October 2010 
    Blogger Comment
    Facebook Comment