Key Nodes Identification Of Wireless Sensor Network Based On Complex Network Theory

As one of the important research branch of complex science, complex network theory has gradually become a hot topic with the rapid development of complex science. Complex network theory is a research tool abstracted from physical networks. It has been widely used in many disciplines, such as biology, chemistry, materials, and other applications. The networks in real world, such as transportation networks, social networks, power grids, and so on are all complex networks. In the research of complex network theory, the key nodes identification method is a very important part and it is significant for maintaining network security. On one hand, we can implement a separate protective measure for these key nodes to enhance network survivability. On the other hand, we can make a deliberate attack to the key nodes, which can lead to the whole network be destroyed. Therefore, the research on key nodes identification in complex networks has important theoretical role in promoting the development of complex network theory, but also has important application value in the real network.Wireless sensor network technology is an important part of internet of things technology. As an entity of complex networks, wireless sensor networks not only has the complex networks’ characteristics, such as large number of nodes, dynamic network and complex network structure, but also have their own unique characteristics, such as the data traffic between the nodes are different and nods are easy to fail because of the harsh environment. It is so important to make sure of the key nodes before distribution, so that can implement individual protective measures, sprinkle individual or adding additional power supply, so as to improve the efficiency, enhance the survivability and extend the life cycle of the wireless sensor network. Thus, the key nodes identification technology for wireless sensor networks has important practical value.This paper studies the key nodes identifications method based on the complex network theory. Aiming at the deficiency of the existing methods in the quantitative of node important extent, we first proposed degree exponent and betweenness exponent to optimize the evaluation of node importance, then by using analytic hierarchy process, combined with the characteristics of wireless sensor networks, we improved the quantitative of nodes weight, the selection of evaluation index and the scientific allocation of index weight and proposed the key nodes identification algorithm in WSNDB-AHP algorithm. The Simulation in classic complex network modes shows that DB-AHP algorithm is more comprehensive and reasonable than only using degree or betweenness or any kind of single method. The main works of this paper include the following contents:First, we have made a detailed analysis of the evaluating nodes importance methods. According to the shortcomings in the social network analysis methods, such as ignoring the whole network’s character, can’t give the node’s important weight a quantized value compare with other nodes or in the whole network. We optimized Degree and Betweenness and proposed the concept of Degree Exponent and Betweenness Exponent.Second, based on degree exponent and betweenness exponent and combined with the characteristics of WSN, this paper applies the AHP in decision theory into complex network and proposes a key nodes identification algorithm in WSN-DB-AHP algorithm. This algorithm makes full use of the advantages of AHP. which combines with qualitative judgment and quantitative analysis. This decisions step makes the final results more scientific and reasonable. DB-AHP algorithm can quantify the nodes weights in the whole network and the weight compared with other nodes. Meanwhile, it considers multiple evaluation indexes as the influencing factors of the importance of nodes, avoiding the one sidedness and not accurate of single method. What’s more, When choosing the indicators, it takes account of the characteristics of wireless sensor network itself, the node energy information is into account, the method is more targeted.Third, we experiment random network model and small-world network mode, which are the most representative models, to verify the actual performance and it shows the universality of the DB-AHP algorithm.Through the experiment, in the random network model, the unreasonable value by using degree method is 20; using betweenness method is 26 and using flow is 4, but the result of DB-AHP is only 1. In the small world network model, the unreasonable value by using degree method is 27; using betweenness method is 27 and using flow is 31, but the result of DB-AHP is 20.No matter what kind of model, the unreasonable value from the DB-AHP algorithm is the smallest. The value in small world network model is bigger than the random network is because of the bigger unreasonable values of degree, betweenness and flow in small world network.Experimental results show that Degree Exponent and Betweenness Exponent proposed in this paper take other nodes’impact to the node into consideration and also be able to quantify the values of the node importance and the values of a node compared to another node in the entire network. To identify key nodes for wireless sensor networks, DB-AHP algorithm is more accurate and effective compared to using a single method or using a variety of methods, but without using Degree Exponent and Betweenness Exponent. The result of identified key nodes is more comprehensive and reasonable. DB-AHP method has good applicability and reliability in wireless sensor networks.