Multilateration Algorithm In WSN Based On K-means Clustering And Data Consistency

Wireless sensor networks (WSN) is a kind of emerging network technology with theadvantages of self-organization, dynamic networking, strong fault tolerance ability,whose functions include information collection, data processing, wireless communication,etc. WSN has already been employed in military, civil, commercial and other areas, andhas been widely used in fire monitoring, medical insurance, earthquake prediction, targettracking, etc. As one of the underlying network technologies that implement the overallperception of the Internet of Things, wireless sensor networks has acquired moreextensive and intensive attention in the era of the Internet of Things.Among most of the applications, node localization is a basic application of wirelesssensor networks. Without location information, any information that wireless sensornetworks monitored would be meaningless.Generally, the process of node localization in wireless sensor networks can bedivided into three steps: information acquisition, location calculation and correction.Multilateration algorithm is widely used in the step of location calculation. Not onlyTOA and RSSI, so called range-based localization algorithms, use multilaterationalgorithm to calculate locations, but also DV-Hop, DV-distance and Amorphous, socalled range-free localization algorithms. Therefore, research on multilaterationalgorithm has great significance in the wireless sensor network localization area.In applications, due to hardware, environmental and many other factors, the distanceinformation that the unknown nodes received may inevitably exist errors. Firstly, thispaper analyzed the sources and approximate ranges of the errors and defined the distanceinformation which exist errors more than five percent of the true distance as inaccuratedistance information. Then, this paper proved that the results of multilaterationalgorithms are vulnerable to the impact of inaccurate information and are restricted to theerrors of the distance information in the last equation.On the basis of analysing and summarizing the existing methods that reduce theinfluence of the errors of distance information on multilateration algorithm, this paper put forwarded a multilateration algorithm based on k-means clustering and dataconsistency. By choosing every three distance information that the unknown nodereceived as a group and then use trilateration algorithm, the algorithm this paper putforwarded can obtain a series of sample points. According to the principle of dataconsistency, the points which calculated with the distance information whose errors arein normal range would be gathered in a dense area whose circle is the real position of theunknown node. While the points which calculated with the distance information whoseerrors are a little large would deviate from this center or even deviate from the dense area.So the points whose errors are a litter large can be picked out by k-means clusteringanalysis method. Then, the distance information whose errors are larger could be foundout and removed so as to cut down the influence of inaccurate distance information.Considering that the result of multilateration algorithm is restricted to the errors of thedistance information in the last equation, the improved algorithm combined every threedistance information and then used trilateration algorithm after removing inaccuratedistance information to elimilate the restriction.Finally, this paper proved that, compared with the original algorithm, the advancedalgorithm can reduce location errors effectively, has stronger fault tolerance ability andmore stable performance without adding any costs of communication by Matlabsimulation experiments in a variety of error environment. It also proved the effectivenessand practicality of the advanced algorithm by simulation experiments based on DV-Hopalgorithm and realistic experiments based on RSSI.