http://bura.brunel.ac.uk/handle/2438/8631 2026-04-09T10:43:03Z 2026-04-09T10:43:03Z Zhang, L Shang, J Wang, Z Liu, Q http://bura.brunel.ac.uk/handle/2438/33115 2026-04-09T02:00:38Z 2026-03-09T00:00:00Z

Title: Remote State Estimation under Stochastic Stealthy Attacks: Short-Term Optimization and Long-Term Convergence Analysis Authors: Zhang, L; Shang, J; Wang, Z; Liu, Q Abstract: This paper investigates the problem of remote state estimation in cyber-physical systems subject to stochastic stealthy attacks. Unlike existing studies that assume persistent intrusion, the attack success is modeled as a stochastic process, thereby providing a more realistic characterization of adversarial capabilities. A comprehensive analysis is conducted from both short-term and long-term perspectives. In the short-term analysis, the evolution of the estimation error covariance is examined, and optimal attack strategies are derived under explicit stealthiness constraints, which limit the detection probability of the attacker. In the long-term analysis, the conditions under which the expected estimation error covariance diverges or converges are explored as a function of the attack success rate and strategy. Rigorous necessary, sufficient, and equivalent conditions for error covariance divergence are established. Moreover, the convergence behavior of the estimation process is characterized under various attack designs, revealing critical thresholds and trade-offs between attack frequency and intensity. Simulation results are provided to validate the theoretical findings and to illustrate the quantitative impact of attack parameters on estimation performance degradation.

2026-03-09T00:00:00Z Song, W Wang, Z Li, Z Dong, H http://bura.brunel.ac.uk/handle/2438/33114 2026-04-09T02:00:36Z 2026-03-11T00:00:00Z

Title: Moving-Horizon Estimation for Multi-Sensor Systems Under Probabilistic Caching Mechanism: A Co-Design Scheme Authors: Song, W; Wang, Z; Li, Z; Dong, H Abstract: In practice, the cache, capable of storing frequently accessed data, is widely deployed in edge servers to guarantee quick retrieval and improve overall system performance. In this paper, the moving-horizon state estimation problem is investigated for a class of multi-sensor systems under the effects of limited caching capacity and sensor resolution. The measurement information collected by multiple sensors is first transmitted to an edge server for state estimation purposes and then stored in the cache for future use. To accommodate the limited caching capacity, the probabilistic caching mechanism (PCM) is harnessed to manage the cached content, under which only a portion of the measurement information is probabilistically selected and retained in the cache. By solving the least-squares optimization problem, a novel moving-horizon state estimator is proposed under the PCM. Sufficient conditions are derived to guarantee that the estimation error is exponentially ultimately bounded in the mean-square sense. To improve the estimation accuracy, the parameters of both the estimator and the PCM are jointly designed by addressing a constrained optimization problem with the assistance of the particle swarm optimization method. Finally, two examples are given to showcase the effectiveness of the proposed algorithm.

2026-03-11T00:00:00Z Chen, Y Wang, Z Wang, J Lan, L http://bura.brunel.ac.uk/handle/2438/33044 2026-03-28T03:00:37Z 2026-02-02T00:00:00Z

Title: Regional Control Subject to Actuator Amplitude and Rate Constraints Under Communication Protocols Authors: Chen, Y; Wang, Z; Wang, J; Lan, L Abstract: This paper addresses the regional control problem for networked systems under simultaneous actuator amplitude and rate constraints. Communication protocols are employed to manage signal transmission over the constrained network. Under the round-robin and try-once-discard protocols, sufficient conditions are derived, expressed as nonlinear matrix inequalities, to ensure the boundedness, H∞ performance, and asymptotic stability of the closed-loop systems. Some algorithms are proposed to optimize the performance indices under linear matrix inequality constraints. Two numerical examples illustrate the validity of the obtained results. Unlike earlier studies that focus mainly on communication protocols or amplitude constraints alone, this paper explicitly incorporates actuator rate constraints and systematically designs static feedback gains rather than assuming them known. As a result, the proposed method yields a less conservative estimate of the domain of attraction even under amplitude-only constraints, while maintaining simpler controller implementation compared to dynamic output-feedback strategies.

2026-02-02T00:00:00Z Jiang, B Wen, J Wang, Z Sheng, W Yu, Z Chen, H Ding, W http://bura.brunel.ac.uk/handle/2438/33043 2026-03-28T03:00:39Z 2026-01-19T00:00:00Z

Title: Scalable Semi-supervised Learning with Discriminative Label Propagation and Correction Authors: Jiang, B; Wen, J; Wang, Z; Sheng, W; Yu, Z; Chen, H; Ding, W Abstract: Semi-supervised learning can leverage both labeled and unlabeled samples simultaneously to improve performance. However, existing methods often present the following issues: (1) The emphasis of learning is put on either the similarity structures or the regression losses of data, neglecting the interaction between them. (2) The similarity structures among boundary samples might be unreliable, which misleads label propagation and impairs the performance of models on out-of-sample data. (3) They often involve the inverses of high-order matrices, making them inefficient in computation. To overcome these issues, we propose a scalable semi-supervised learning framework with Discriminative Label Propagation and Correction (DLPC), which collaboratively exploits the regression losses and similarity structures of data. Particularly, each sample is projected onto the independent class labels associated with nonnegative adjustment vectors rather than the propagated labels, such that the distances between samples from different classes are naturally enlarged, making regression losses more effective for boundary samples. Benefiting from this, the regression losses can guide the propagation of labels in boundary areas. Thus, the label information is first propagated through dynamically optimized graph structures and then corrected by the regression losses, effectively improving the quality of labels and facilitating feature projection learning. Furthermore, an accelerated solution has been developed to reduce the computational costs of DLPC on sample scales, thereby making it scalable to relatively large-scale problems. Moreover, the proposed DLPC can not only be applied to single-view scenarios but also extended to multi-view tasks. Additionally, an optimization strategy with fast convergence has been presented for DLPC, and extensive experiments demonstrate the effectiveness and superiority of DLPC over state-of-the-art competitors.

2026-01-19T00:00:00Z