Cyber-security is of the fundamental importance for cyber-physical systems (CPSs), since CPSs are vulnerable to cyber attack. In order to make the defensive measures better, one needs to understand the behavior from the view of an attacker. In this paper, the problem of false data injection attack on remote state estimation with resource constraints is studied in two cases, where the first case is that the attacker adds a Gaussian noise to the innovation, while the other is that the attacker employs a Gaussian noise to replace the innovation. In addition, the attacker is assumed to has a resource constraint, i.e., he/she cannot attack all the sensors, at the same time should decide which sensors to attack. By using the matrix theory, the optimal attack strategy problem, which aims to maximize the trace of the remote estimation error covariance, is converted into a convex optimization problem that can be solved. Thus, an optimal attack strategy is given to illustrate which sensors should be attacked. An example is given to show the effectiveness of the theoretical results.
24 February 2017 in IEEE Transactions on Neural Networks and Learning Systems
This paper investigates the problem of output feedback adaptive stabilization for a class of nonstrict-feedback stochastic nonlinear systems with both unknown backlashlike hysteresis and unknown control directions. A new linear state transformation is applied to the original system, and then, control design for the new system becomes feasible. By combining the neural network's (NN's) parameterization, variable separation technique, and Nussbaum gain function method, an input-driven observer-based adaptive NN control scheme, which involves only one parameter to be updated, is developed for such systems. All closed-loop signals are bounded in probability and the error signals remain semiglobally bounded in the fourth moment (or mean square). Finally, the effectiveness and the applicability of the proposed control design are verified by two simulation examples.
7 December 2018 in IEEE Transactions on Neural Networks and Learning Systems
Probabilistic Boolean network (PBN) is a kind of stochastic logical system in which update functions are randomly selected from a set of candidate Boolean functions according to a prescribed probability distribution at each time step. In this brief, a pinning controller design algorithm is proposed to set stabilize any PBN with probability one. First, an algorithm is given to change the columns of its transition matrix. Then, according to the newly obtained transition matrix, a fraction of nodes can be selected as pinning nodes to inject control inputs to achieve set stabilization. The problem is challenging since the Boolean functions in a PBN are not deterministic but are randomly chosen among several Boolean functions. Furthermore, the structure matrices of the pinning controllers are given by solving some logical matrices equations based on which a pinning controller design algorithm is provided to set stabilize the PBN with probability one. Finally, the theoretical results are validated using several examples.
16 July 2020 in IEEE Transactions on Industrial Informatics
In this paper, a defense method with watermarking to detect linear deception attack under K-L divergence detector in Cyber-physical system (CPS) is proposed. It is known that linear deception attacks can reduce the performance of remote estimator without being detected by the K-L divergence detector. In order to detect this kind of attack, we use watermarking to encrypt and decrypt data transmitted through wireless networks. When the attack does not exist, the transmitted data can be restored to ensure remote estimation performance. In the presence of linear deception attacks, these data are marked with a watermarking so that they can assist the K-L divergence detector to discover the attack. The watermarking encryption method is proved to be helpful for K-L divergence detector to discover attack, or weaken the impact of the attack in different situations. Finally, numerical simulations are provided to further illustrate the results.
16 December 2019 in IEEE Transactions on Industrial Informatics
In this article, we consider the security problem for the consensus-based distributed state estimation. To resist the malicious attacker who can falsify the data transmitted through the wireless channel, each node equips with an attack defender, which is based on the measurement of its built-in sensor. Under the stealthy attack, which can deceive the defender, we investigate the resilience and convergence of the distributed estimation in two different attack scenarios. For the attack with enough communication resources, we provide a sufficient condition of the optimal attack to quantify the maximum estimation performance degradation. We also analyze the resilience of the worst case distributed estimation caused by the attacker. For the attack with limited resources, the optimal Kalman gain for each node is derived to maximize its estimation performance under the attack. We also give a sufficient condition to guarantee the convergence of the distributed estimation in this case. Finally, numerical simulations are provided to illustrate the effect of the defender on guaranteeing the resilience of sensor networks against attacks.