Research On Acquisition And Application Of Location Information Based On Wireless Sensor Networks

Node position is the foundation of various applications on wireless sensor network,so the problem of how to obtain the location accurately and securely become a hot topic of current WSN research.Moreover,under the condition that the localization error is inevitable,it is necessary to study the associated strategy which has strong anti-interference capability and good adaptability to ensure the subsequent execution of related application.After analyzing research of existing work,the thesis mainly focused on three issues,including node localization,localization security and geographic routing resilient to location errors based on employing statistical,convex optimization,pattern recognition theory,etc.The main work and achievements are specially shown as follow:(1)Two localization methods,RS-PG and AIWO-Powell,are proposed respectively,which solve the problem that the RSSI based static localization is susceptible to the influence of ranging error.According to shadowing attenuation in wireless channel,RS-PG algorithm divides the intersection region that is covered jointly by beacons into several small grids,and calculates the value of likelihood function to estimate the coordinate of unknown node,which reduces the influence of noise on localization accuracy;In order to deal with the scene that anchors are deployed at the fringe of networks,AIWO-Powell algorithm utilizes an improved invasive weed optimization algorithm to solve the objective function of localization.Simulation studies show that these two new algorithms always outperform other localization algorithms for positioning precision,and perform efficiently even in high range error scenario.(2)A clustering and multidimensional scaling based mobile-assisted localization approach(CMM)is proposed for nonuniform deployed wireless sensor networks.The scheme firstly splits the network into several clusters in order to perform the positioning task in a distributed fashion;and then a novel path planning method for mobile beacon is proposed by using artificial potential field theory which can effectively reduce the movement distance;Finally,a nystr?m approximated multidimensional scaling method is presented to estimate the node location,which reduces the amount of computation from the cluster with high node density.Experimental results show that CMM method can greatly improve the performance of positioning accuracy,and has better anti-interference ability.(3)An alternating direction method of multipliers--p-norm Support Vector Machine(ADMPSVM)recognition algorithm is proposed,which overcomes the drawback that the traditional secure localization algorithm cannot identify the type of external localization attacks.The approach first introduces p-norm regularization into support vector machine,which enhances the classification and generalization ability for various kinds of attack;Then it combines the pnorm SVM classifier with ADMM optimization method to assign the computation task to adjacent nodes,and nodes cooperate with each other to obtain the recognition result.The simulation results validate the ADM-PSVM algorithm not only realize the accurate recognition for diverse types of external localization attack,but also has better computational efficiency.(4)An exchange based attack classification algorithm(PECPR-MKSVM)is proposed according to the fact that original feature is insufficient to classify the type of external localization attack.This algorithm can improve the distinguishing ability between features and computational efficiency by modeling mixture distribution,extracting statistical features and optimizing the distributed support vector machines.Simulation results show the proposed method can improve the recognition accuracy and operation efficiency significantly.(5)A geographic routing for RSSI based weighted centroid localization condition is proposed,The method solves the problem that existing geographic routing against location error took no account the practical influence of the specific localization technology on positioning error when it is applied in wireless sensor networks.Based on the model of RSSI measurement,this method first explores the statistical characteristics with respect to error of forwarding distance after weighted centroid localization,and it use Nakagami distribution to describe the multipathfading status more precisely.And then a new localization error-resilient geographic routing is proposed,which regards the minimum rms error of distance expectation as main delivery basis combined with node residual energy and deviation angle.Simulation results show that accounting for error influence of specific localization significantly reduces the failure of packet delivery ratio as well as energy consumption.