Research On Relative Localization Algorithm With Multidimensional Scaling Technology On Wireless Sensor Networks

With the rapid development of embedded computing and wireless communications technology,wireless sensor networks which represents the next generation of Internet of Things technology has been the most popular study field in information technology.However,localization technology,the key supporting technology in the wireless sensor networks,has been the high attention by various aspects and so far a lot of technic programs and algorithms had been proposed but most of them had limited precision,high cost of hardware or difficulty to realize.Therefore,it has a high theoretical and practical value to study and design a localization algorithm which is easier to apply on mainstream hardware platforms and meets the application needs based on existing wireless sensor networks.In this paper,based on the research and analysis of a large number of wireless sensor networks localization algorithms,especially relative localization algorithms,an algorithm based on measurement information from hardware called HIB-MDS(Hardware Information Based-Multidimensional Scaling)has been proposed which is feasible and convenient to apply in reality.On this basis,a clustering relative localization algorithm based on tree structure called TBC-MDS(Tree Based Cluster-Multidimensional Scaling)has been further proposed in order to improve the scalability of the HIB-MDS and its coverage and accuracy.The major researches and contributions include:1.Wireless sensor networks and its localization technology have been introduced.A large number of research papers about wireless sensor networks localization,particularly relative localization,has been read and introduced with the analysis of their advantages and disadvantages.2.The theory of multidimensional scaling technology has been elaborated,two MDS technologies-metric based and nonmetric based-have been introduced and how these technologies are used in wireless sensor networks localization has been explained.Based on the research on relative localization of wireless sensor networks combined with current technology,application status,and prior researches,a relative localization algorithm based on measurement information(RSSI&LQI)from hardware called HIB-MDS has been proposed.Its feasibility and advantage has been verified in two ways:simulation experiments using MATLAB and real experiments using GAINSJ node packages.3.A structured clustering method based on the tree structure called TBC clustering algorithm has been proposed.Using HIB-MDS localization algorithm as in-cluster localization algorithm,a clustering relative localization algorithm based on tree structure called TBC-MDS has been designed and its ability and contribution to localization coverage and accuracy improvement has been verified in MATLAB simulation experiments.The innovation of this study is that the hardware measurement information has been used as the formation data of dissimilarity matrix in MDS technology to reduce the systematic errors produced by data conversion.Structured clustering method has been used in the cluster formation stage which is to ensure the number of public nodes between clusters can meet the minimum requirements for the merging stage,and control the degree of overlap between the clusters.At the same time,scheduling problem in cluster merging stage has been effectively solved with the reduction of errors produced by merge stage.Various experiments show that compared to traditional MDS localization algorithm the proposed HIB-MDS relative localization algorithm has higher localization accuracy,by which localization error rate is close to or even lower than 15%.In real conditioned experiments,the absolute values of localization error remain within the range from 1.5 to 4 meters.At the same time,the coverage rate of proposed TBC-MDS clustering relative localization algorithm can achieve more than 90%,or even completely cover the whole network in certain conditions.The localization error rate of TBC-MDS algorithm has a better performance than traditional distributed MDS algorithm in that the rate can be controlled under 20%or below for the best and will not be higher than 40%in low network connectivity.