Research On Source Location Privacy Security Protection In Cyber-physical Systems

Cyber-physical systems(CPS)refering to a new generation of systems with integrated computational and physical capabilities which involve large-scale areas that provide unprecedented performance and efficiency levels through sophisticated control-computing codesign.CPS have aroused concerns of governments,business,and academia around the world.With the rapid progress,CPS will inevitably result in high potential security and privacy threats.CPS are the networked systems with feedback mechanisms essentially.The networked is the major feature which distinguish them from other control systems.Thus,it is of both practical and theoretical significance to study the CPS security and privacy from the perspective of network.Particularly,the location of the event source could be possibly exposed to the attacker.Since the attacker always monitors the network,which can easily eavesdrop the message from the path between the source and the sink.To address the source location privacy(SLP),the corresponding researches include mechanisms of flooding,random walk,fake source,etc.Although great progress has been made in this field,there still exist many shortcomings for protecting the CPS,which have the characteristics of limited resource,real-time response,high fault tolerance.In view of these shortcomings,this master's thesis studies attacker location evaluation-based fake source scheduling,limited-sampling active learning tracing attack,differential private autonomous underwater vehicle.The specific research contents in this thesis are as follows:1.For the trade-offs between source location privacy,transmission delay,and energy consumption,the attacker location evaluation-based fake source scheduling(FSSE)is proposed.The proposed FSSE mainly contains two phases.One is namely backbone construction which to build a backbone to form a baseline in respect of the privacy and the transmission delay while considering a captured probability derived from the communication information of self and neighboring nodes.The other is namely fake message scheduling which is established to determine nodes to be fake sources in terms of the hypothesis location of the attacker by using stochastic process and the state of self.2.For the problem of the unreasonable random walk attack model,the limited-sampling active learning tracing attack is proposed to form an intelligent attack model for evaluating SLP in CPS.The limited-sampling attack model is established while considering the restriction of filtered and labeled samples caused by limited system knowledge.The network traffic classifier is designed to optimize tracing strategy while considering the infinite tracing steps may occur under the unrestricted random walk.The active learning using the combination of two representation and diversity of samples is proposed to solve the problem of the unbalanced dataset and outliers on account of major or minor fake network traffic.3.For the problem of the source device privacy in CPS,a proportional-integral-derivative(PID)autonomous underwater vehicle(AUV)is investigated,while considering a network attacker eavesdrops and gains the message from the communication channel between the AUV and the data center.The differential private noise-based data upload mechanism is designed to defend against the network attacker.Considering differential private protection,the Kalman filter-based trajectory evaluating attack is designed to analyze the trajectory privacy.Then,the particle swarm optimization-based parameter evaluating attack is considered to analyze the parameter privacy of the PID controller.Finally,simulations show that the algorithms proposed in this paper are efficient.The fake source mechanism is established.Privacy level is evaluated from the perspective of attacker.