19 April 2023 in IEEE Transactions on Information Forensics and Security
The security problem of state estimation plays a critical role in monitoring and managing operation of cyber-physical systems (CPS). This paper considers the problem of network security under replay attacks and formulates a novel attack detection method. More specifically, we design a model-free reinforcement learning-based replay attack detection framework that can automatically learn and recognize the evolving attacks with more effectiveness. Attackers in some situations are more like intelligent agents with initiative, who can transform their attack strategies purposefully according to the actions of defenders. Thus, we propose a new defense strategy against the interaction between the attacker and the defender which is solved by optimization learning. The proposed analytical procedure concerning reinforcement learning technology can also be extended to the study of other control applications. Finally, the numerical examples are provided to illustrate the effectiveness of the detection method.
29 April 2022 in IEEE Transactions on Automatic Control
This article studies the security of distributed state estimation under data integrity attacks. Each sensor is equipped with a Kullback–Leibler (K–L) divergence detector to diagnose the authenticity of the received data. A necessary and sufficient condition for the insecurity of the distributed estimator is derived, in which the stealthy attack can evade the detector and degrade the estimation performance. Furthermore, an algorithm is proposed to generate stealthy attack sequences. To overcome the vulnerability, a data transmission strategy based on watermarking is proposed. The effect of the watermarking parameters on the attack in different scenarios is subsequently analyzed. It is proved that the K–L divergence detector can effectively distinguish the stealthy attack with the help of the proposed strategy, thereby ensuring the security of the distributed state estimator. Compared with traditional encryption mechanisms, the proposed strategy neither increases the computational complexity nor sacrifices the estimation performance in the absence of attacks. A simulation example is presented to demonstrate the effectiveness of the developed approach.
16 May 2022 in IEEE Transactions on Information Forensics and Security
This paper studies the defense strategy of remote state estimation under deception attacks. In order to prevent the stealthy attacker from reducing the estimation performance without triggering an alarm, an encoding-decoding mechanism combining linear transformation and artificial noise is proposed. Moreover, the detection performance under three different attack scenarios is analyzed. It is proved that the false data detector can effectively identify the attack or weaken its impact on the system under the proposed strategy, so as to ensure the security of the system. From the perspective of an attacker, an algorithm that can deduce the approximate values of the encoding parameters is also provided, which reveals how the magnitude of the artificial noise affects the accuracy of the attacker’s inference. Finally, a simulation example is presented to verify the effectiveness of the developed approach.
10 November 2020 in IEEE Transactions on Automatic Control
This work addresses reset moving horizon estimation for multiple output discrete-time systems with quantized measurements. A new state reset estimator is designed based on a one-dimension noisy measurement to overcome underestimation or over-estimation of the system state, and an iterative algorithm is proposed to deal with multiple output systems. It is shown that with the proposed reset algorithm, the state estimation error is improved in the presence of over or under estimation, and the boundedness of the estimation error is established. The proposed algorithm also achieves a better estimate than the existing one for systems with a scalar measurement in the static case. A simulation of a moving vehicle is provided to demonstrate the advantage of the developed approach.
28 February 2022 in IEEE Transactions on Control of Network Systems
Average consensus is a major operation in distributed Kalman filtering. It requires neighboring nodes to exchange state information with each other, which may result in undesirable private data leakage. Since distributed Kalman filtering requires that the estimate at each time instant is accurate, it brings more challenges to design privacy-preserving scheme for operation. In this article, we design a privacy-preserving scheme for distributed Kalman filtering without the loss of estimation performance, which is also suitable for average consensus or dynamic average consensus of multiagent systems. We first build a secure multihop communication based on an encryption scheme. We then calculate the sum of the states of neighboring nodes with secure summation, which ensures that the state update will not reveal the state of the node to its neighboring nodes. We employ different methods to calculate the sum of the states of neighboring nodes against noncollusive and collusive adversaries. For the noncollusive case, the privacy of the honest nodes is preserved. For the collusive case, if there are too many adversaries, the privacy of the honest nodes could be exposed when accurate distributed Kalman filtering is accomplished. Therefore, we measure the risk of the global system suffering privacy leakage as the privacy index and improve the ability of the system to defend the collusive adversaries by using a group-based method. Some numerical examples are provided to illustrate the effectiveness of the proposed schemes.
6 April 2023 in IEEE Transactions on Information Forensics and Security
This paper studies the confidentiality issue of wireless sensor networks under communication constraints from the perspective of an eavesdropper. The considered communication constraints include both medium access constraint and bandwidth limitation. Specifically, the former results in only one sink node gaining access to the shared network at each time step, while the latter requires data to be quantized by a probabilistic quantizer before being transmitted. The task of the eavesdropper is to develop eavesdropping strategies for the constrained network to infer the state of the system as accurately as possible. Since the data is generally unreadable to unauthorized third parties in terms of secure transmission, an eavesdropper with limited power must decide which sink nodes’ data needs to be decrypted. By analyzing the impact of different decryption strategies on the eavesdropping performance, a deciphering scheduling is proposed, which minimizes the expected estimation error without exceeding the energy budget. Besides, a recursive reset algorithm is put forward based on the properties of the probabilistic quantizer, which reflects that an eavesdropper can infer whether its own estimate is accurate enough from the decoded data. Moreover, a sufficient condition is established in which the eavesdropping performance is improved under the proposed algorithm. A numerical example is provided to demonstrate the validity of the developed approaches.