Research On Distributed State Estimation And Detection Of Wireless Sensor Networks Under Malicious Attacks

With the rapid development of digitization,informatization and intelligence,Wireless sensor network plays an irreplaceable role in unmanned driving,intelligent transportation,healthcare system and other aspects.However,due to the vulnerability of network,communication process may suffer malicious attacks,so an authenticity and integrity of datas will be destroyed.In addition,it is difficult to use the standard Kalman filter method to accurately estimate state of systems because most systems in practical environment are nonlinear systems.Therefore,it is significant to study distributed secure state estimation of nonlinear systems under malicious attacks.The specific research content is as follows,Firstly,a distributed secure state estimator with random switching topologies based on D-S evidence theory detection is proposed to deal with possible false data injection attacks in interaction process of smart sensors.The detector takes signal-tonoise ratio,state estimation and measurement into consideration,improves traditional D-S fusion rules,and reduces influence of attacks on the estimator.Considering the influence of neighbor state prediction information on estimation effect,a distributed estimator of double gains based on detection algorithm is designed.In addition,sufficient conditions are given for the estimation error covariance to be bounded.Secondly,aiming at more hidden and destructive man-in-the-middle attacks,a distributed consensus on information secure state estimation algorithm based on RSA encryption and Hash function verification is designed.The detection mechanism uses RSA encryption to improve data confidentiality and Hash function to detect man-inthe-middle attacks that crack passwords to improve data security.Based on the above detection result,a distributed Kalman-consensus filtering algorithm is proposed to ensure consensus of estimation state,and it is proved that information matrix is bounded under given sufficient conditions.Finally,considering that RSA encryption algorithm costs a lot of resources and distributed consensus on information estimation algorithm cannot integrate neighbor information,a distributed consensus on information and consensus on measurement secure state estimator based on elliptic curve encryption and Hash function detection algorithm is founded.The designed detector not only greatly reduces resource consumption,but also improves security of systems.On this basis,the improved estimator can integrate neighbor measurements to avoid correlation assumption between local estimates.In addition,convergence of estimation errors of the designed estimator is analyzed under given sufficient conditions.