Research On Privacy Protection And Accuracy Of Localization In The Internet Of Things

Location information is one of the most fundamental elements for the Location-Based Services(LBSs)in Internet of Things(IoT)environments since it is an important feature of the physical world.Localization has been an enabling technology for the communications and services which are dependent on the location of the things and users in IoT.Therefore,it is significant to research on IoT Localization.Even though the advanced satellite-based radionavigation systems,such as the Global Positioning System(GPS),are able to provide localization services,they do not work in non-line-of-sight environments,such as indoor environments,which leads to the inappropriateness for IoT applications.As a result,it is necessary to comprehensively study localization in order to enable the LBS applications for the resource-constrained mobile devices in IoT environments.Existing work has made many achievements and progress for the research on localization.However,there still are two important problems faced by localization in IoT environments.One aspect is the shortcoming of the privacy protection schemes during the localization process,which may cause private location information leakage.The other aspect is the limitation of the localization error controlling schemes,which may introduce inaccuracy and high-cost for the localization system leading to be unable to meet the requirements of users.Therefore,in terms of both theoretical and application aspects,it is of great significance to develop localization algorithms so as to achieve privacy-protecting and high-accurate localization technologies.To address the above challenges,this dissertation investigates the privacy-protecting and high-accurate localization methods for IoT environments.In order to achieve privacy preservation,anti-cyber-attacks,high accuracy and low cost for IoT localization systems,we propose a privacy-protecting and high-accurate localization solution,which includes efficient privacy-preserving localization,anti-colluding-attack based anchor node selection,accurate location region estimation,and single-anchor localization.The major contributions of this dissertation are summarized as follows.Firstly,efficient privacy-preserving localization is investigated in order to utilize no homomorphic encryption.An adjacent subtraction-based localization(SAL)model is first introduced through analyzing the traditional nonadjacent subtraction-based localization(NSL)model.According to the SAL model,an efficient privacy-preserving localization(EPPL)algorithm is developed by applying the three privacy-preserving building blocks,i.e.,privacy-preserving summation,privacy-preserving adjacent product summation,and privacy-preserving adjacent difference summation.Then,the SAL model is analyzed by proving the equivalence between SAL and NSL models in terms of the localization error.According to a bounded error model,the lower and upper bounds of the localization error are derived.The correctness and privacy preservation of the EPPL algorithm are analyzed and proved.The communication and computational overheads of the EPPL algorithm are also analyzed.According to the results of real-world experiment platforms,simulations are conducted to illustrate the equivalence between ASL and NSL,and the performance of EPPL regarding the correctness,privacy,and efficiency.Then,anti-colluding-attack based anchor node selection is studied for localization to efficiently select anchor nodes and defeat colluding attacks.A novel capability-induced colluding attack model is first proposed through analyzing the structure of the colluding attacks among the attackers,where attackers with different capability levels are more likely to collude with each other.In order to both solve the colluding attack problem and improve the efficiency of the anchor node(participant)selection,a resilient participant selection(RPS)algorithm is developed to select participants in polynomial time,by deriving the sufficient condition and the necessary condition of achieving the colluding possibility minimization among participants.Then,it is proved that the participants selected by the RPS algorithm are able to perform all the tasks and achieve the minimized colluding possibility.In terms of the task-performing cost and the potential damage caused by the colluding attacks,it is proved that the RPS algorithm has a smaller social cost than the existing algorithm of minimizing the task-performing cost.Real-world-trace-based simulations are conducted to demonstrate the effectiveness of RPS and the correctness of the theoretical results.Moreover,accurate location region estimation is explored to control the localization error.In order to accurately estimate the location region of the target node,the location region estimation problem is investigated using the distance distribution theory.Through modeling the target node as the random node inside a disk region,a novel disk error model is first presented to capture the uncertainties during the distance ranging process.According to the disk error model,a disk-error-based ranging(DEBR)method is proposed by estimating the parameters of the disk error model.By taking into account both DEBR and the classical multi-lateration method,an accurate location region estimation(ALRE)algorithm is developed to calculate the location region of the target node.Then,the correctness of the DEBR method is analyzed by proving that the parameter estimation of DEBR is unbiased.The relationship between the localization error and the number of the anchor nodes is analyzed through deriving the closed-form expression of the multi-lateration method.It is proved that the estimated region obtained by ALRE is tighter than that obtained by the traditional estimation method.Simulations are conducted to verify the unbiased estimation of DEBR and evaluate the performance of ALRE.Finally,single-anchor localization is modeled and applied to improve the performance of traditional localization algorithms.In order to perform high-accurate localization using a small number of anchor nodes,the single-anchor localization is utilized to increase the accuracy of the traditional multi-lateration method.The single-anchor localization has been an important research topic since a multi-antenna anchor node is able to estimate the location of the target node using both angle and distance information.The single-anchor localization model is first presented to describe the localization process using a single anchor node.Based on the single-anchor localization model,an accurate and distributed localization(ADL)algorithm is proposed,which can not only estimate the location of the target node with fewer anchor nodes but also be more accurate than the traditional multi-lateration method.Then,it is proved that the location estimate under ADL can converge towards the real location of the target node with probability 1.The lower and upper bounds of ADL are also derived under a bounded noise model.Simulations are conducted to demonstrate the performance of ADL and the correctness of the theoretical results.