Research On Secure Localization Methods For Wireless Sensor Networks

Wireless Sensor Network(WSN)is a hot research which integrates sensor technology,embedded computing technology,microelectronics technology and wireless communication technology.It is commonly used in many fields,such as military defense,environmental monitoring,biomedicine,traffic management,smart home,agricultural production and so on.It is considered to be one of the most important technologies in the 21st century.Determining the location of node or event plays a key role in the effectiveness of its application,and is also the premise and foundation for the construction and maintenance of the sensor network,event monitoring,target tracking and other functions.However,due to the limited resources of the node,the openness of the deployment area and the broadcasting characteristics of radio itself,the positioning process has security risks.Therefore,how to ensure the location of node safely with the presence of security threats remains a fundamental problem to be solved,which has important theoretical significance and application value.This paper focuses on the research of security localization methods for wireless sensor networks.Starting from different application scenarios,on the basis of in-depth study of related technologies at home and abroad,the paper focuses on solving the difficulties and challenges in the field,including the localization attack classification for wireless sensor networks,robust security localization for static and dynamic wireless sensor networks,and position verification for the wireless body area networks.The main work of this paper includes:(1)In order to address the problem of localization attack classification in wireless sensor network,a localization attack classification method based on deep learning architecture is proposed.On the base of the original location information,the method introduces the node degree and clustering coefficient indicator based on the complex network theory into the feature representation according to the change rule of the network topology under attacked and non-attacked.The Stacked Denoising Auto-encoder(SDAE)is used to carry out unsupervised pretraining layer by layer,and maps a large number of unlabeled high-dimensional data into an optimal low-dimensional representation.The low-dimensional representation of the SDAE out-put is then classified as supervised.The proposed approach could efficiently distinguish normal beacons and five kinds of malicious beacon nodes launch the Sybil attacks,Replay attacks,Interference attacks,Collusion attacks,Wormhole attacks.The experimental results show that the localization attack classification algorithm using SDAE is superior to the deep learning model using the Stacked Auto-encoder and SVM in recognition time and recognition rate.(2)In order to solve the problem that the malicious beacon node in the static wireless sensor network can provide the wrong position information to affect the positioning accuracy and validity,a method which can not only tolerate the misleading information provided by the malicious beacon node,but also quickly and accurately determine the location information of the unknown node is proposed.Firstly,distance information between the beacon node and the unknown node is obtained by the classical distance location algorithm.The maximum likelihood position estimation probability model is constructed by maximizing distance error probability density.To solve the problem that the designed model is a nonlinear minimization,a simple and effective gradient descent algorithm is utilized to find the optimal solution to meet the requirements of wireless sensor network security localization.Due to the defect of randomly initial value,a trilateral centroid method based on 3 far-neighbors is proposed to narrow the search range of the position estimation and improve the localization accuracy.As some large distance residual vectors in distance residual gradients produced by malicious attacked beacon nodes and the localization accuracy affected the measurement noise,a secure localization method based on the weighted robust gradient descent is proposed.More secure and accurate localization is achieved by combining the suspiciousness weight that can weaken the impact of the inconsistent malicious distance residual vectors with the distance weight between nodes that can improve ranging accuracy.Simulation data show the proposed algorithm is superior to the related security localization algorithms in runtime,security and robustness.(3)In order to solve the secure localization problem where beacon nodes in mobile wireless sensor networks are unavailable,a robust security localization method for mobile wireless sensor networks,based on the proposed network security localization method for static wireless sensor,is proposed.It can tolerate malicious position reference attacks from untrusted neighbor nodes,with the help of the relative location map.The relative distances between nodes and neighbor relationship are saved in the relative location map.If the absolute positions of any three nodes are known,the absolute positions of the remaining nodes can be obtained by the conversion parameters.Simulation results show that the proposed method is more resistant to malicious attacks and achieve high localization accuracy.(4)In order to solve the problem that the fact,the physical position of the sensor node placed on the patient's limbs may mismatch the preset position in the wireless body area network,leads to the risk of miscalculating the vital signs of the patient,a sensor node position detection method for application security of the wireless body area network is proposed.In order to reduce node resource overhead,increase node universality,and reduce the participation of patients in identification,the barometric altimetry technology with the RSSI positioning technology based on the received signal strength indication are combined.It can automatically identify and map which sensor device is placed on which human limb based on atmospheric pressure value and RSSI information,without the participation of beacon node.Double sample T test is used to quantify the reliability of the method to ensure the correct recognition of 95%confidence.When sharp fluctuations in short-term pressure values caused by extreme weather may affect the differentiation of upper and lower limb sensor,the height of sensor nodes can be estimated by a differential pressure altimetry to eliminate the influence of weather environment.Gaussian filtering is used to weaken the influence of noise.Experiments indicate that the proposed method could easily recognize a 40-cm horizontal body range and a 65-cm vertical body range.It is suitable for identifying sensor positions in an indoor environment.Compared with other indoor positioning methods,the method does not need to learn a large amount of data in advance,and achieves certain adaptability even if there are a large number of environmental factors in the indoor environment.