Distributed Filtering Algorithms For Sensor Networks Under False Data Injection Attacks

Distributed sensor networks are widely used in various fields of civil and national defense,such as environment monitoring,robot collaborative work and UAV formation flying,etc.In a distributed sensor network,due to the interconnections between different components,as well as the interconnections between sensors and other agents,the transmitted data is vulnerable to malicious attacks,which can cause the destruction of the sensor network.In order to effectively resist the impact of external malicious data injection attacks,this paper designs distributed filtering algorithms with attack detection,which can effectively identify false information and maintain the system's secure operation.Moreover,considering the introduction of attack detectors will greatly increase the energy consumption of sensor nodes,this paper proposes an event-triggered communication mechanism in the sensor network to limit the transmission of some unimportant data,so as to reduce the energy consumption of the sensors.In summary,this paper focuses on studying injection attacks and energy limitation and designing distributed filtering algorithms to achieve distributed secure estimation of the target while reducing energy consumption of sensor nodes as much as possible.The detailed contents are as follows:1.Measurement attack detection algorithm and distributed filter design for distributed sensor networks.For sensor networks with intermittent measurement attacks and a target with input and measurement noise,this thesis designs an attack detection based distributed filtering algorithm to improve the distributed filtering performance.By studying the measurement residuals between the prior measurement data and the actual measurement data of sensor nodes,the detection conditions of the sensor node under injection attack are obtained,so as to remove the attacked data,design consensus based information filtering algorithm,and then ensure the normal operation of the system.2.Design of distributed filtering algorithm with measurement attack detection and eventtriggered communication mechanism.On the one hand,a measurement attack detection mechanism is introduced to eliminate the abnormal data under attack.On the other hand,on the basis of attack detection,a consensus Kalman filtering algorithm based on event-triggered communication mechanism is proposed.Nodes only transmit their secure local information to neighboring nodes when the most recently propagated data deviates from the current data,which thereby reduces the communication bandwidth in the network improves energy efficiency,and prolongs network life.3.Transmission channel attack detection algorithm and distributed filter design for distributed sensor networks.By performing the backup update on the latest data transmitted by the neighbor node that was judged to have no attack,the difference between the received data from the neighbor and the backup update data can be obtained to detect the attack.In addition,the data fusion flag is established.When the backup data of the neighbor node is excessively deviated,the use of the backup data of the neighbor node will be abandoned in this data fusion and local data will be used for replacement.Then,based on the attack detection mechanism,a consensus based information filtering algorithm is proposed.