Maneuvering target tracking in three-dimensional space remains a challenging problem due to complex motion dynamics and model mismatch. To address this, this paper proposes a hybrid model/data-driven algorithm named IMMNet, which integrates the interpretable structure of the interacting multiple model (IMM) algorithm with learnable neural components. Unlike end-to-end black-box methods, the proposed IMMNet algorithm not only can preserve the Bayesian inference mechanism that is essential for real-time radar applications, but also can adaptively learn motion patterns and noise characteristics from data. Extensive experiments demonstrate that the proposed IMMNet algorithm consistently outperforms the existing algorithms across various scenarios, validating it as a robust, interpretable, and practical solution for maneuvering target tracking.

This letter addresses real-time single-UAV path planning in obstacle-rich environments and proposes an ET-PSO-APF. An enhanced artificial potential field provides fast local guidance, while a risk index with hysteresis and a refractory period triggers PSO-based local replanning only when reliability degrades. Once activated, PSO searches within an adaptive window and horizon; a fallback ensures safety if optimization stalls. Simulations across sparse, normal, and complex fields show ET-PSO-APF yields shorter, smoother, and safer trajectories with modest runtime compared with H-PSOAPF, PSO, and APF algorithm.

Multi-target tracking (MTT) in real-world environments often faces the challenge of outlier measurements, which severely degrades the performance of standard MTT algorithms. This letter integrates the maximum correntropy criterion (MCC) into the Gaussian mixture probability hypothesis density (GM-PHD) filter, an implicit data association and a highly efficient MTT algorithm. The MCC provides a localized similarity measure that is inherently resilient to impulsive outliers. We embed an iterative fixed-point measurement update for the GM-PHD filter, and an adaptive kernel size design strategy is also devised. The proposed MCC-GM-PHD filter effectively suppresses the influence of large measurement residuals, while maintaining a closed-form Gaussian mixture representation. The performance of the proposed MCC-GM-PHD filter is verified via simulations.

This article addresses the problem of tracking a time-varying number of multiple direction-of-arrival (multi-DOA) by exploiting the received signals of a coprime array. The multi-DOA state is represented by a labeled random finite set, which is further modeled using the generalized labeled multi-Bernoulli (GLMB) distribution. Therefore, the GLMB filter naturally arises as a suitable method to propagate the multi-DOA posterior, from which the multi-DOA state can be extracted. To capture the intrinsic high-dimensional characteristics of the signals received by the coprime array, a tensor model is adopted, followed by the canonical polyadic decomposition to extract the rank-1 components of the tensor, which contain the spatial information corresponding to each source. Building on such a result, we formulate the multi-DOA likelihood in the form of a von Mises-Fisher distribution. This likelihood enables the derivation of the update equation for the GLMB filter. Furthermore, we developed a particle-based implementation of the algorithm, and its effectiveness is subsequently assessed through simulation experiments. Even under low signal-to-noise ratio and limited snapshot conditions, the proposed algorithm can accurately track the multi-DOA of time-varying source numbers, with the generalized optimal subpattern assignment error remaining within 0.5(degrees)

Adversarial sensing is a security related problem in cyber physical systems (CPS), where a defender controlled target may be tracked by a non-cooperative adversary. The inverse filters, which are launched on the defender, have emerged as tools to infer the exposure of defender’s state to the adversary, which employs a standard filter (referred to as forward filter in this letter), such as Kalman filter, to recursively estimate the defender’s state. This letter proposes an inverse variational Kalman filter (IVKF), which extends the inverse Kalman filter to unknown and even time-varying adversarial measurement noise covariance (MNC) that plays an essential role in the forward (i.e., standard) filtering algorithms. The idea is to jointly estimate the inverse state and adversarial MNC, while the MNC cannot be straightforwardly estimated. Thus, the variational Bayesian (VB) framework is leveraged, and the stochastic natural gradient VB (SNGVB) inference is derived so as to solve the problem of non-conjugacy in computing the posterior of MNC. The performance of proposed IVKF is validated through simulations.

This paper presents an efficient implementation of the extended object Poisson multi-Bernoulli (PMB) filter under the zero-inflated Poisson (ZIP) object measurement model using particle belief propagation (BP). The ZIP measurement model separates a Bernoulli object detection event from the conditional Poisson generation of object measurements, enabling principled handling of empty measurement sets. Building upon the PMB mixture posterior, we present a factorized joint posterior over set of objects with object detection variables and a dual representation of data association using both object-oriented and measurement-oriented association variables. Notably, this representation replaces the implicit high-order global hypothesis constraint by local consistency factors, yielding a factor graph amenable to BP. In addition, we present a particle-based implementation, where the single object densities of Bernoulli components are represented using particles. Simulation results show that the proposed method achieves filtering performance comparable to a sampling-based PMBM implementation, while having lower runtime. We also validate the efficacy of the proposed method using real-world lidar data for pedestrian tracking.

Information dissemination in dynamic social networks enables fast and frequent access to social news. Thereinto, the coexistence of public and private information creates a hybrid dissemination dynamics process in social networks. However, most existing information dissemination models treat the hybrid information in isolation and fail to consider their interactions through shared nodes and temporal dependencies. Thus, we propose a novel hybrid information dissemination (HID) model that explicitly captures the interconnected dissemination mechanisms of both public and private information within dynamic social networks. Additionally, considering heterogeneity among individuals, we further design a decision-making algorithm for the proposed HID model, aiming at maximizing individuals’ initiative. Furthermore, we derive equilibrium points and analyze their stability for the proposed HID model. Numerous experiments are conducted, and results show that the proposed HID model can effectively describe the dissemination process of hybrid information.

This paper addresses the problem of distributed multitarget tracking, where sensors exhibit heterogeneous resolution capabilities. Such a scenario arises in applications such as mixed pedestrian and vehicle tracking with different sensor types, where the vehicle may occupy only a few resolution bins in some sensors, leading to a point target representation, while it may be detected across multiple resolution bins in other sensors, resulting in an extended target representation. The number of occupied resolution bins may also vary over time due to changes in the target's distance from the sensors, occlusions, or environmental conditions. To tackle this challenge, we represent the multitarget state using a labeled random finite set (LRFS) and model it with the labeled multi-Bernoulli (LMB) density. An adaptive LMB filter is proposed to dynamically adjust the identity of each target (point or extended target) based on the available sensor measurements. Specifically, when a target is observed with only a few measurements, it is treated as a point target due to insufficient resolution for shape estimation. Conversely, when high-resolution measurements are available, the target's extended state, including shape and orientation, is estimated. To account for the dynamic nature of target appearances, we integrate a Markov transition model into the LMB filter, allowing targets to switch between point and extended representations over time. We also introduce a computationally efficient implementation of the proposed filter based on Gibbs sampling. Furthermore, a distributed fusion strategy is developed to aggregate local LMB posteriors from multiple sensors, enhancing tracking accuracy in large-scale sensor networks. The proposed algorithms are evaluated based on simulations concerning distributed tracking of pedestrians and vehicles, and the KITTI dataset and V2V4Real dataset are also employed to examine the effectiveness of proposed algorithms on real applications. All these results demonstrate the effectiveness of the proposed algorithm, showcasing its robustness against varying sensor resolutions and its applicability to next-generation intelligent transportation systems.

PRIMEX (prime-based graph encoding and extraction) is a recently proposed framework for scalable distributed fusion. In PRIMEX, the information pedigree of state estimates or probability density functions is encoded using the information codes, enabling lightweight arithmetic for redundancy removal and data integration. Building on PRIMEX and its memoryless fusion strategy based on a least-squares approximation, in this paper we present two efficient distributed tracking algorithms: a consensus-based PRIMEX method that fuses information from all neighbors, and a greedy gossip-based PRIMEX method that fuses with the most informative neighbor. To further increase communication efficiency, we incorporate an event-triggered mechanism, in which transmission decisions are driven by information novelty measured using differences between the information codes. The proposed methods are evaluated and compared with covariance intersection and centralized fusion in a distributed single target tracking scenario. Simulation results show that PRIMEX-based methods remain competitive in tracking accuracy while improving communication efficiency.

The significant differences in the size, contour shape, and structural features of extended targets, as well as the dynamic changes under different sensor perspectives, pose significant challenges to extended target tracking (ETT). However, a general ETT framework that balances shape estimation accuracy and model complexity is still lacking. Thus, based on Gaussian process theory, this study investigates the adaptive adjustment of measurement model according to the target’s shape information distribution, and proposes a measurement-driven ETT algorithm. The proposed algorithm utilizes the distribution characteristics of real-time measurement density to capture the target’s shape information, and designs a strategy to dynamically adjust the number and positions of basis points, which more effectively represents the complex target shape. Simulation results show that the proposed algorithm significantly improves the estimation accuracy of the extended target by only using 1/3 of the number of basis points compared to existing algorithms, providing an efficient solution for ETT to balance the trade-off between accuracy and complexity.

Unmanned Aerial Vehicles (UAVs) are renowned for their agility. As UAV swarms take on increasingly complex collaborative tasks, they face a growing threat from Byzantine attacks, where compromised UAVs transmit misleading information, affecting missions and leading to failures. Unlike traditional attacks, Byzantine attacks on UAV swarms occur during the online learning of coordination, where real-time model updates are critical. Existing research largely overlooks this dynamic and evolving vulnerability. This paper proposes a novel trustworthy hierarchical aggregation algorithm within a federated reinforcement learning framework, specifically designed to defend against Byzantine attacks by ensuring robust and accurate aggregation of model parameters during the online learning process. This resilient aggregation mechanism significantly enhances the swarm's learning integrity in adversarial environments. We design a realistic UAV round-up scenario. Simulations demonstrate that UAV swarms with our algorithm maintain strong resilience, even under severe Byzantine attacks, highlighting its promise for secure and reliable UAV swarm intelligence.

This paper considers the multi-target tracking (MTT) problem under epistemic uncertainty, and such a goal is achieved by integrating possibility theory into the Poisson multi-Bernoulli mixture (PMBM) filtering framework. To do so, we first define the possibility PMBM, and then we derive the possibility PMBM filtering recursions. The resulting possibility PMBM filter preserves strong theoretical foundations of PMBM while enhancing robustness to model mismatches. In addition, we present the possibility Poisson multi-Bernoulli (PMB) filter, which is a computationally efficient approximation of the possibility PMBM filter. We also present analytical implementations of the proposed possibility PMBM and possibility PMB filters based on Gaussian mixture representation and their robustness and estimation accuracy have been demonstrated in the simulation studies.

The vulnerability of wireless communication links to eavesdropping poses significant challenges in securing AAV-assisted networks. To enhance security, reconfigurable intelligent surfaces (RIS) and artificial noise (AN) have emerged as promising technologies for mitigating eavesdropping by controlling wireless propagation environments and introducing interference against eavesdroppers. However, existing works have rarely combined transmitter beamforming, RIS, and AN integratedly considered, and leveraging their complementary characteristics for efficient security enhancement remains challenging. Additionally, optimizing such system security performance is complicated by the nonconvexity of secrecy rate maximization and the highly time-varying communication links caused by the mobility of AAVs and users. To address these challenges, we propose a secure communication framework that integrates RIS and AN transmission devices on AAVs. To solve the resulting nonconvex optimization problem, we develop a dual-mode framework based on twin delayed deep deterministic policy gradient (TD3), employing two subenvironments that interact independently before updating a global environment. Extensive simulations demonstrate that the proposed approach significantly enhances secrecy rate performance compared to other methods.

As UAV technology has become more popular in various fields, tracking UAV swarms has gradually become a hot topic in recent years. Traditional tracking algorithms struggle to distinguish individual targets within the swarm under conditions of high clutter rates, low measurement rates and limited sensor resolution, resulting in limited available measurements. In this scenario, tracking each target within the swarm becomes impractical, necessitating a focus on the overall state of the UAV swarms rather than the individual states of its constituents. However, the existing methods have shown unsatisfactory tracking performance for UAV swarm in challenging scenarios with low measurement rates. To enhance the tracking performance in such challenging scenarios, in this paper, we propose a UAV swarm tracking method, called ST-UST, which combines a deep learning network named swin transformer with the Kalman filter. The core element of the proposed ST-UST method lies in the utilization of swin transformer to process images derived from noisy point clouds. In this method, swin transformer can achieve the inference of swarm shape parameters, and the Kalman filter is utilized to estimate swarm kinematic parameters. Experimental results show that, in comparison with the existing methods, the proposed ST-UST method has significant competitiveness in challenging scenarios with low measurement rates, and the maximum tracking accuracy can be improved by nearly 20.7%.

This paper presents an order allocation algorithm incorporating dynamic information to improve the allocation plan for drone order delivery. The algorithm utilizes the Hungarian algorithm to obtain an initial allocation plan and adjusts it based on factors such as the distance between new orders and drones, as well as drone idle time. Path-planning algorithms are designed to handle obstacles, represented as circles in the real world. The proposed methods are evaluated using the Zhuoyi simulation platform, and extensive experiments conducted in three scenarios demonstrate their superior performance compared to the two baseline approaches.

A mixed-criticality system integrates subsystems with different criticality levels on the shared hardware platform, significantly improving resource utilization and becoming an important research topic for embedded systems. However, most existing works on scheduling strategies for mixed-criticality systems are based on independent task sets. This paper designs a scheduling strategy considering task data dependencies and real-time operating system constraints. In addition, this paper saves energy costs by using DVFS to adjust task execution frequency dynamically. Specifically, a multiple DAG task scheduling strategy is designed for the heterogeneous platform, exploring idle time through DVFS to optimize system energy cost in low-criticality mode with the consideration of ensuring task schedulability while fully utilizing multiprocessing to enhance system performance in the high-criticality mode. Experimental results show that our method can guarantee the quality of service and optimize the system energy cost compared with the existing methods.

Collision-free path planning is a significant challenge in deploying uncrewed aerial vehicle (UAV) swarms for sensor data collection tasks, particularly in uncertain environments. This article addresses the problem of designing flight strategies for energy-constrained UAV swarms using deep reinforcement learning (DRL). We propose a locally aware double prepartitioned deep Q-network (DPPDQN) for flight planning of multiple UAVs in large-scale uncertain industrial environments. This algorithm aims to collect data from ground sensors with communication constraints while avoiding mutual collisions. In the proposed algorithm, we first introduce a node prepartitioning scheme using an improved clustering algorithm, which efficiently reduces the number of access points for UAVs through effective data aggregation. Subsequently, we employ an area prepartitioning algorithm that segments the detection area into several subregions based on the number of UAVs available for operation, thereby preventing collisions among the UAV swarm during task execution. Finally, we propose a deep Q-Network (DQN) algorithm incorporating a convolutional neural network (CNN), which utilizes the local perceptions of UAVs as state inputs to address the "curse of dimensionality" occurring in large-scale environments. Systematic simulation experiments validate the effectiveness and advantages of the proposed method.

Dear Editor, This letter focuses on how an attacker can design suitable improved zero-dynamics(ZD)attack signal based on state estimates of target system.Improved ZD attack is to change zero dynamic gain matrix of attack signal to a matrix with determinant greater than 1.

Extended target tracking in occlusion scenarios often suffers from split errors due to sensor limitations and complex target interactions, leading to degraded tracking performance for autonomous vehicles and surveillance systems. To address this issue, in this paper, we propose a Gaussian Wasserstein distance-enhanced spatio-temporal similarity method for split error correction. We first analyze the spatio-temporal characteristics of split extended targets and model their geometric uncertainties via elliptical Gaussian distributions. Then, we integrate the Gaussian Wasserstein distance into the clue-aware trajectory similarity calculation framework to simultaneously capture positional and shape discrepancies, and designs an adaptive validation gate mechanism to dynamically adjust the threshold for track splitting, enabling accurate determination and fusion of split targets. Finally, simulation experiments are conducted to demonstrate the effectiveness of the proposed method.

This paper investigates unmanned aerial vehicle (UAV) path planning for data collection from ground-constrained sensors in large-scale industrial environments. Initially, clustering of nodes is performed to reduce the number of data access points. To address issues such as improper setting of cluster heads and data loss during data transmission among sensor nodes, a Dynamic Cluster Head Selection algorithm based on K-Medoids (DSK-Medoids) is proposed in this paper. This algorithm effectively tackles communication quality issues and enhances data collection. Additionally, the introduction of Deep Q Network (DQN) for path planning overcomes the “curse of dimensionality” problem associated with traditional path planning algorithms. This enables intelligent path planning strategies and improves UAV flight paths and data transmission efficiency. Through systematic simulation experiments and performance analysis, the effectiveness and performance advantages of the proposed methods are validated.