Research On Security Estimation Based On Wireless Sensor Network Under Deception Attack

With the continuous development of wireless communication,embedded computing technology,sensor technology and micro electro mechanical technology,wireless sensor network(WSN)has changed the way of interaction between human and nature,showing great application value in military,environmental monitoring,medical care,target tracking,etc.It can not only realize the collection and arrangement of data,but also realize the highspeed transmission of data,which can greatly improve the work efficiency.Because WSN is mostly deployed in the open environment,it is difficult to avoid the external environment or potential security threats.Deception attack is an important means used by network attackers,which can destroy the data transmission process and even make WSN paralyzed..In addition,due to the complexity of the perceived object,the system itself often has unknown input,state saturation and random nonlinearity,which undoubtedly increases the difficulty of security mechanism and estimation algorithm design.In this paper,aiming at the above problems in WSN,the network attack modeling,security mechanism and estimation algorithm design are studied.The specific research contents are summarized as follows:Firstly,for a class of linear systems with unknown input,the design of minimum variance estimation algorithm is studied when the wireless channel suffers from false data injection(FDI)attack.The unknown input is treated as a process of unknown prior information,the attack signal is modeled as a random deviation,and the detector is used to judge whether the wireless channel is attacked.According to the detection results of the detector,the time series can be divided into two types.Under the two types of time series,the minimum variance state estimation algorithm is designed,the convergence and stability of the algorithm are further analyzed.Secondly,considering the situation of multi-sensor measurement,we study the design of defense strategy and distributed estimation algorithm when WSN suffers from false data injection attack.For each sensor node,the residual formed by the state estimation information from the node itself and the neighbor is used as the evaluation function.By comparing the evaluation function with the preset threshold,the sensor can broadcast data to the network conditionally,thus reducing the impact of the tampered data on the estimation effect.An event-based distributed security estimation algorithm is designed.The estimator gain is obtained by minimizing the upper bound of estimation variance,and the sufficient conditions are given to ensure the asymptotic boundedness of estimation error covariance.Finally,for a class of state-saturated multi-sensor network system with stochastic nonlinearity,the problem of attack modeling,defense strategy and distributed estimation algorithm design is studied when sensor nodes suffer from deception attack.In order to reduce the influence of deception attack on estimation effect,an attack detection mechanism with adaptive threshold is proposed,and the performance of the detection mechanism is analyzed,the threshold condition for detecting deception attack completely is given.Furthermore,a distributed security estimation algorithm is designed,and the sufficient conditions to ensure the boundedness of mean square of estimation error are given.