This article investigates the problem of partial node-based (PNB) recursive state estimation for complex networks (CNs) with unknown nonlinearities and energy harvesting sensors. To mitigate the effects of energy constraints, an energy replenishment mechanism is employed, in which a group of energy harvesting sensors captures energy from the surrounding environment. These sensors transmit measurement outputs to remote state estimators only when their current energy levels are sufficient to cover the transmission energy costs. By exploiting the universal approximation property, neural networks (NNs) are utilized to approximate the unknown nonlinearities of the CNs. An NN-based recursive estimation algorithm is developed to simultaneously generate the estimates of the system state and the unknown nonlinearities. Following a specific set of recursions, the recursive state estimator gains and the NN weight (NNW) tuning parameters are calculated in a unified framework. Finally, the effectiveness of the developed recursive estimation algorithm is demonstrated through a simulation example.

This paper investigates the HPo observer design problem for Internet of Things (IoT)-enabled time-delay systems subject to periodic disturbances and bandwidth-limited communications. Motivated by practical IoT scenarios in which spatially distributed sensor nodes share constrained wireless channels, a random access protocol is adopted to regulate data transmissions, allowing only one sensor node to access the network at each time instant. Two disturbance structures are examined: (i) periodic disturbances entering only the system state, and (ii) periodic disturbances simultaneously affecting both the system state and the measurement output. For each case, an iterative learning observer (ILO) is developed to jointly estimate the system state and periodic disturbances across repeated operation cycles. The proposed ILOs ensure that the state estimation error satisfies a prescribed HPo performance level, despite the combined influences of communication-induced intermittency and periodic disturbances. Sufficient conditions for the existence of the desired ILOs are derived in terms of matrix inequalities, from which the observer gain matrices are computed. Simulation studies demonstrate that the proposed IoToriented observer designs achieve accurate disturbance reconstruction and resilient estimation performance under random access communication constraints.

The federated-filtering-based (FFB) fusion estimation problem is investigated in this paper for networked multi-rate systems, where the measurement signals are transmitted over a wireless network with limited transmission power. A probabilistic quantization mechanism is introduced to handle the raw measurement signals for the purpose of facilitating digital communication over network. Certain transmission models are proposed to describe the behaviors under the effects of multi-rate dynamics, probabilistic quantization and limited transmission power. A delicately designed FFB fusion scheme is proposed to acquire the desired state estimates, where the local filters will receive feedback from the fusion center to reset their estimates. The parameters for the local filters are calculated by recursively minimizing their upper-bounds for the estimation error covariances. Furthermore, new conditions have been derived to analyze the ultimately boundedness of the estimation error covariance for the fusion center. Subsequently, a power allocation strategy is designed by minimizing such ultimate bound subject to the given transmission power constraint. Finally, the effectiveness of the proposed fusion estimation strategy and its optimal power allocation scheme is verified through a simulation example. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

Time series forecasting is crucial in many real-world fields. Most state-of-the-art forecasting models now use the Transformer architecture, which has succeeded by effectively modeling long-range dependencies. However, its quadratic computational complexity hinders practical deployment, especially for long-sequence tasks, limiting its use in resource-constrained real-world scenarios. Both the state space model (Mamba) and 1D Convolutional Neural Networks (1D-CNN) achieve linear computational complexity. Mamba excels at capturing global dependencies, while 1D-CNN effectively captures local patterns, and their complementary strengths enable efficient modeling across different temporal scales. In this paper, we propose a novel model, A Reverse Multi-Scale Time Series Forecasting Method with Mamba and 1D-CNN (RMTSFMC). First, a reverse encoding mechanism is introduced to process multi-scale time series data, and then an effective feature extractor is designed. Within this extractor, the 1D-CNN layer supplements local information from large to small scales, and the multi-scale residual Mamba layer extracts global information. The two complement each other and efficiently capture time series features. The constructed future prediction module fuses and outputs multi-scale features to realize the effective integration of information at different scales. In the experiments, we evaluate our proposed method on seven common benchmark datasets. Compared with the baseline model TimeMixer, the Mean Squared Error (MSE) and Mean Absolute Error (MAE) on the Traffic, ETTh1 and ETTh2 datasets are improved by 4.21%, 4.57%, and 4.40% (for MSE) and 3.39%, 2.24%, and 2.44% (for MAE) respectively, and comparable performance is achieved on other datasets. This indicates that the model achieves good generalization performance and improved prediction accuracy. While maintaining performance, it improves efficiency and reduces memory usage with linear computational complexity.

The Internet of Things (IoT) increasingly relies on large-scale networked agents that cooperate over bandwidth-limited and unreliable wireless links. This paper studies the finite-horizon H∞ consensus control problem for a class of discrete-time IoT-enabled multi-agent systems subject to random parameters, stochastic communication scheduling, and relay-assisted transmissions. To reflect practical IoT communication constraints, a novel transmission framework is considered that integrates a stochastic communication protocol with a decode-and-forward relay mechanism under random packet losses. Within this framework, only a subset of sensing information is scheduled for transmission, encoded, decoded, and forwarded by relay nodes to improve communication reliability and coverage. Based on the resulting networked system model, a distributed consensus controller is designed to attenuate the effects of disturbances and communication uncertainties over a finite time horizon. Sufficient conditions guaranteeing finite-horizon H∞ consensus are derived in terms of recursive linear matrix inequalities, and a corresponding controller synthesis algorithm is developed. Simulation results illustrate the effectiveness of the proposed approach in achieving robust consensus for IoT-enabled multi-agent systems operating over stochastic relay-assisted communication networks.

In this paper, the problem of resilient recursive state estimation is addressed for a class of nonlinear cyber-physical systems operating under token bucket protocols and subject to probabilistic bit flips. Measurement signals are transmitted to the remote estimator only when the token storage surpasses the token consumption required for transmission. The communication process employs a binary encoding scheme, which quantizes measurement outputs into a bit string, transmits them through memoryless binary symmetric channels subject to probabilistic bit flips, and subsequently recovers them at the receiver. To achieve the desired estimation performance, a resilient state estimator is developed to mitigate the adverse effects of random perturbations in the estimator gain during implementation. The aim is to design a recursive state estimation algorithm that effectively manages the token bucket protocol, addresses probabilistic bit flips, and accommodates estimator gain perturbations. An upper bound for the estimation error covariance is derived, and the corresponding estimator gain is recursively calculated to minimize this bound. Finally, numerical simulations are conducted to validate the effectiveness of the proposed algorithm. (c) 2026 Published by Elsevier Ltd.

In this paper, the secure set-membership state estimation problem is investigated for a class of networked linear systems, where the measurement data might be intercepted by potential eavesdroppers. To protect the privacy of system state from information leakage, an artificial-noise-assisted encryptor is dedicatedly designed to transform the measurement data into the ciphertext (i.e., the encrypted data) before being transmitted, and a decryptor is then employed at the state estimator side to decrypt the received ciphertext. Under the proposed encryption-decryption mechanism, the concept of secrecy capacity is introduced to quantify the information security of the signal transmission process. A parameter-dependent state estimator is constructed to confine the estimation error into a time-varying ellipsoidal set. The desired parameters for the state estimator and the encryptor are co-designed by resorting to a set of recursions. Furthermore, sufficient conditions are derived to guarantee the ultimate boundedness of the time-varying ellipsoidal set. Finally, two simulation examples are provided to demonstrate the effectiveness of our developed secure set-membership state estimation scheme. (c) 2025 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

In this paper, we investigate the fault estimation problem for stochastic system with filter-and-forward successive relay channel. The successive relay mechanism employs two half-duplex relay nodes that alternately receive and transmit mea- surement from the sensor to the remote estimator, respectively. This scheme introduces switching characteristics and inter-relay self-interference. The primary objectives of this article are to design relaying strategy and the remote estimator subject to the comprehensive effect induced by the switching behavior of relaying mechanism, inter-relay self-interference and the stochas- tic characteristics of the system. Specifically, we ensure that, at each sampling instant, the estimation error covariance has an upper bound that is minimized by appropriate design of the estimator gain. Furthermore, sufficient conditions are provided guaranteeing the boundedness of the proposed estimator. Finally, a numerical example is provided to demonstrate the effectiveness of the obtained results.

This article addresses the problem of input-output data-based ultimate boundedness control for a class of networked systems subject to probabilistic bit flips and false data injection (FDI) attacks under the try-once-discard (TOD) protocol. First, a prior experiment is conducted to obtain a set of input-output data from the considered system, which will be utilized for the data-based controller design. A uniform-quantization-based encoding-decoding mechanism is employed to digitalize measurement signals. The TOD protocol is adopted to schedule signal transmissions between encoders and decoders. Considering the nature of digital communication, an ellipsoid constraint and a sequence of Bernoulli variables are introduced to account for the FDI attacks and bit flips during the transmission, respectively. To expediently design the data-based controller, a novel autoregression (AR)-based method is proposed, subject to probabilistic bit flips and protocol-induced effects. This article aims to design a data-driven controller that ensures the ultimate boundedness of the closed-loop system under the effects of TOD protocol scheduling and communication failures. Sufficient conditions are presented to ensure the desired control performance by using the S-Lemma from data. An improved cone complementarity linearization (CCL) algorithm is developed to calculate the controller gain. Eventually, a numerical simulation example is provided to demonstrate the effectiveness and feasibility of the proposed data-based ultimate boundedness control scheme.

In this article, the H infinity fuzzy control problem is investigated for a class of nonlinear systems subject to replay attacks with frequency-duration constraints. Owing to the vulnerability of the open shared communication network, the information transmitted from the sensor to the controller may be exposed to replay attackers. A novel yet comprehensive replay attack model is constructed to characterize the repeated replay behavior of the adversary. On the basis of the constructed model, a fuzzy controller is designed to guarantee asymptotic stability and the desired H infinity performance. By employing Lyapunov stability theory and the orthogonal decomposition technique, sufficient conditions are derived to ensure the existence of the desired controller parameter. Finally, simulation results are presented to verify the effectiveness and correctness of the proposed fuzzy controller for Takagi-Sugeno (T-S) fuzzy systems under replay attacks.

This article addresses the problem of distributed state estimation for smooth nonlinear systems over sensor networks by means of a generalized fuzzy proportional-integral observer (PIO). A sensor network is employed to collect system measurements, with a pull-type gossip protocol governing the intermittent data exchange among neighboring nodes. Under the gossip protocol, each sensor node randomly selects one neighbor to request data, facilitating distributed information updating. Furthermore, considering challenges, such as long-distance communication and complex environmental conditions, signal transmission is subject to amplitude fading. To accommodate the characteristics of the gossip protocol, a generalized fuzzy PIO with a flexible structure is developed. Sufficient conditions are derived to guarantee the H-infinity estimation performance of the proposed observer. Based on established conditions, the parameters of both the gossip protocol and the fuzzy PIO are codesigned via a particle-swarm-optimization-based iterative algorithm, with emphasis on enhancing observer robustness. Finally, an engineering-oriented simulation example is presented to illustrate the effectiveness of the proposed methodology.

The control problem of nonlinear systems under network transmission constraints has become a central research topic in networked control systems (NCSs). While communication networks introduce flexibility and scalability, their inherent limitations disrupt the information conditions and structural assumptions required by traditional nonlinear control methods, thereby posing significant challenges to system stability and performance. These issues have led to extensive efforts to develop nonlinear robust control frameworks tailored for networked environments. This survey reviews the main progress in this area by examining three representative classes of nonlinear systems with distinct structural features: strict-feedback/pure-feedback systems, nonlinear systems with matched uncertainties, and feedback-linearizable systems. The specific ways in which network-induced phenomena destroy the structural premises of these control approaches are clarified, including interruptions of recursive chains in backstepping, violations of arrival and sliding conditions in sliding mode control, and distortions of state transformations in feedback linearization. The solution strategies proposed to address these challenges are then grouped and summarized according to their underlying mechanisms. Finally, key conclusions are drawn, and several potential research directions are outlined to guide future work on nonlinear control under network transmission constraints.

This paper is concerned with the fault estimation problem for discrete time-varying systems subject to loss of control effectiveness over relay-aided communication channels. Under effects of relay-aided communication, the signal transmission between the plant and the remote fault estimator is implemented over three communication processes (namely, the sensor-to-estimator, the sensor-to-relay and the relay-to-estimator communication). To facilitate digital communication, several groups of encoders and decoders are employed for the three communication processes, where bit flips might occur due to the effects of channel noises. Three Bernoulli-distributed random variables are used to describe the occurrences of such bit flips. To restrain the adverse impact induced by bit flips, a novel decode-and-forward relaying technique is developed, where the forwarded signal is determined by a delicately designed event. Two coupled difference equations are proposed to characterize the upper-bound of the corresponding estimation error covariance. Then, the desired estimator gain matrix is calculated via minimizing the resultant upper-bound. Finally, a numerical example is provided to verify the effectiveness of our proposed relaying technique and fault estimation scheme.

This article is concerned with the recursive state estimation issue for a class of nonlinear cyber-physical systems (CPSs) with token bucket protocols (TBPs) subject to sensor failures and false data injection (FDI) attacks. In the system under consideration, measurement signals are transmitted to the remote estimator only when there are sufficient tokens in the bucket to meet the token consumption. During network transmissions, the signals are exposed to FDI attacks, which occur randomly and follow a Bernoulli distribution. The primary objective is to develop a state estimation algorithm that can handle the TBP, sensor failures, and FDI attacks simultaneously. Initially, the upper bound of the EEC is derived using an intensive stochastic technique and the induction approach. Subsequently, the desired estimator gains are recursively computed to minimize this upper bound. Finally, an example is presented to demonstrate the effectiveness of the proposed estimation scheme.

This article addresses the problem of secure recursive state estimation for a networked linear system, which may be vulnerable to interception of transmitted measurement data by eavesdroppers. To effectively protect information security, an encryption-decryption-based communication scheme can be used, but encrypting all the measurement data from sensors can result in significant computational costs. To address this issue, a partial-encryption-decryption (PED) mechanism is proposed to enhance information security with relatively low computational costs. In this mechanism, only part of the transmitted measurement signals are encrypted, and the remaining signals are transmitted directly to the estimator. A Jordan-canonical-form-based approach is developed to select the appropriate parameter for the PED mechanism, and recursive formulas for the state estimator are designed based on the principle of minimum mean squared error. Sufficient conditions are derived to guarantee the ultimate boundedness of the estimation error variance matrix. Finally, the proposed PED-based recursive state estimation scheme is evaluated through two simulation examples to demonstrate its effectiveness.

This article investigates the proportional-integral-derivative (PID) containment control problem for a class of linear MAS with multirate measurements under the constraint of sensor resolution. The sensors of agents are classified into two distinct groups, characterized by their relatively fast and slow sampling periods. The concept of sensor resolution is introduced to quantify the ability of sensors to detect the smallest changes in information. A PID controller with an improved structure is proposed to achieve containment control, ensuring that follower agents remain within the convex hull formed by the leader agents. The closed-loop system is reformulated into a simplified representation, incorporating both sampling characteristics and communication topology. Sufficient conditions are then derived to guarantee the exponentially ultimate boundedness of the tracking error. Based on these conditions, an iterative algorithm is developed for computing the required controller gains. Finally, a simulation study, along with comparative analyses, is conducted to validate the effectiveness of the proposed approach.

This article addresses the issue of robust finite-horizon H-infinity filtering for complex networks subject to replay attacks. A replay attack strategy is implemented by the adversary on the communication channel between the network nodes and the filters, with the intention of replacing the current measurement data with previously recorded measurement data. Considering the limited energy of the attacker, a binary variable is adopted to indicate whether the communication channel is under attack. To better characterize the dynamic behavior of replay attacks, two factors dependent on attack frequency and a time-varying parameter are introduced. Subsequently, under the impact of replay attacks, the switched filtering error dynamics is obtained with a time-varying delay. By employing the average dwell-time method, sufficient conditions are derived to guarantee the weighted H-infinity performance of the filtering error dynamics. Furthermore, the filter gain parameters are computed through the solution of some recursive matrix inequalities. Finally, numerical simulation results are conducted to verify that the developed filter design algorithm is effective.

This work addresses the problem of recursive state estimation for networked control systems with unknown nonlinearities and binary-encoding mechanisms (BEMs). To enhance transmission reliability and reduce network resource consumption, BEMs are used to convert measurement signals into binary bit strings (BBSs) of limited length, which are then transmitted to the estimator through noisy communication channels. During transmission, random bit errors may occur in the BBSs due to channel noise. For the considered nonlinear networked control systems affected by random bit errors, a neural-network (NN)-based recursive estimation strategy is proposed, where an NN with a time-varying tuning scalar is employed to approximate the unknown nonlinearity of the networked control systems. By using the proposed strategy, the upper bounds of the estimation error of the system state and the trace of the estimation error of the NN weight (NNW) are first derived. These bounds are then minimized by recursively designing both the estimator gain matrix and the tuning scalar of the NNW. Finally, the effectiveness of the proposed estimation strategy is demonstrated through a numerical example.

This paper presents a cross-domain robot initial localization method based on Siamese neural networks with parameter-efficient adapters. Traditional methods such as Adaptive Monte Carlo Localization(AMCL) and Point-to-Line Iterative Closest Point (PL-ICP) struggle to achieve reliable initial pose estimation in dynamic environments. Our method performs a location matching by learning discriminative embeddings of LiDAR scan data. Lightweight adapter modules enable rapid cross-environment adaptation while preserving spatial reasoning capabilities. Experimental results demonstrate that the method achieves outstanding performance with only a small number of samples from the target environment. Compared to traditional methods, the proposed approach exhibits higher accuracy and stronger environmental adaptability.

In this paper, a novel covert attack detector is developed for cyber-physical systems by using input-associated watermarking mechanism. Covert attacks are very sophisticated and rely on perfect model knowledge, so that the generated attack signals have no response in the measurement output and thus evades the standard chi(2) detector. A necessary and sufficient condition and a sufficient condition are respectively presented for the stealthiness and destructiveness of covert attacks for the purpose of revealing the features of covert attacks. Then, a novel input-associated watermarking mechanism is developed to assist chi(2) detector to "expose" the attack behaviors in the detection process. Intensive analysis is subsequently implemented on the resultant attack detection rate under the proposed detection mechanism. It is worth emphasizing that the developed detection mechanism does not sacrifice the estimation performance under the attack-free situation. Furthermore, the designed proactive detection method is extended to the detection for the so-called zero dynamics attacks. Ultimately, a simulation example is utilized to illustrative the usefulness and effectiveness of the proposed input-associated watermarking mechanism. (c) 2025 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.