Research On The Mds-based Localization Algorithm Of Wireless Sensor Networks

Wireless sensor networks is a wireless network system consisting of largenumbers of tiny sensors which is capable of calculation and communication andcontact in multi-hop self-organizing way.Wireless sensor networks has brought a newmode of information acquisition and information processing,and it has widespreadapplication in military,surrounding monitoring, calamity prediction and businessmanagement etc.Location information of nodes is very important in the relatedapplications of WSN.Sensor nodes must know its own position to specify a particularevent where it happen and realize positioning and tracking of external objective;on theother hand, the location information of sensor nodes can also improve the efficiencyof routing,provides network naming space,report coverage quality of network to thedeployer,balance load and realize the self-configuration of network topology.We caneasy obtain the nodes’ location by manually setting the node position or install GPSchip,but it will take more time and resources, so using localization algorithm toestimate the nodes’ position is a better approach.Multidimensional Scaling technology (MDS),derived from psychometri cs,is akind of common data analysis technology to express multidimensional datarelationship as geometry figure and study it.Since2003,Shang Y et al applied MDS toWSN for the first time and proposed MDS-MAP localization algorithm,MDS wasincreasingly used to solve the problem of node location,and got new progressconstantly. MDS uses connectivity and distances information of the network to createsimilar matrix,and transforms this matrix to get the nodes’ relative coordinates.In allusion to centralized MDS-MAP localization algorithm has largecommunication consumption leading to early death of the nodes near the centernode,high computational complexity and error accumulation of hops,this paperproposes a fast clustering algorithm for WSN-adaptive and searching algorithm.Itrealize distributed positioning by dividing the entire network into many intersectingthe local map.Clustering can also improve the scalability and robustness of thealgorithm.Because the nodes’ distribution is relatively uniform in a cluster,it is also available for large-scale irregular network.For classical MDS using the shortest path distance to replace the actualdistance of nodes,which leads to the similar matrix has a big error,an improvedQMDS-MAP (Quadrilateral-based MDS-MAP) localization algorithm is proposed inthis paper.It combines offset refinement with Euclidean algorithm to calculate thedistance between nodes,but no longer uses the shortest path distance.In order toimprove the positioning accuracy of QMDS-MAP,it use the constant modulusalgorithm (CMA) to optimize the local map after obtaining the local relativecoordinates.And then merging them into global relative map and transforming it toglobal absolute map by the known anchor nodes.Thesis carries out research on the localization problem of WSN,and focuses onthe MDS-based localization algorithm,and proposes QMDS-MAP on thosebasis.Through the MATLAB simulation and comparison with MDS-MAP andEMDS-MAP (P),results show that the proposed algorithm can effectively improve thepositioning accuracy,reduce the computational complexity and communicationconsumption，and have good adaptability for the large-scale sparse network andlarge-scale irregular network.