The vision of the upcoming 6G technologies that have fast data rate, low latency, and ultra-dense network, draws great attentions to the Internet of Vehicles (IoV) and Vehicle-to-Everything (V2X) communication for intelligent transportation systems. There is an urgent need for distributed machine learning techniques that can take advantages of massive interconnected networks with explosive amount of heterogeneous data generated at the network edge. In this study, a two-layer federated learning model is proposed to take advantages of the distributed end-edge-cloud architecture typical in 6G environment, and to achieve a more efficient and more accurate learning while ensuring data privacy protection and reducing communication overheads. A novel multi-layer heterogeneous model selection and aggregation scheme is designed as a part of the federated learning process to better utilize the local and global contexts of individual vehicles and road side units (RSUs) in 6G supported vehicular networks. This context-aware distributed learning mechanism is then developed and applied to address intelligent object detection, which is one of the most critical challenges in modern intelligent transportation systems with autonomous vehicles. Evaluation results showed that the proposed method, which demonstrates a higher learning accuracy with better precision, recall and F1 score, outperforms other state-of-the-art methods under 6G network configuration by achieving faster convergence, and scales better with larger numbers of RSUs involved in the learning process.

Smart manufacturing aims to support highly customizable production processes. Therefore, the associated machine intelligence needs to be quickly adaptable to new products, processes, and applications with limited training data while preserving data privacy. In this article, a new federated transfer learning framework, federated transfer learning for cross-domain prediction, is proposed to address the challenges of data scarcity and data privacy faced by most machine learning approaches in modern smart manufacturing with cross-domain applications. The framework architecture consists of a central server and several groups of smart devices, where each group handles a different application. The existing applications can share their knowledge through the central server as base models, while new applications can convert a base model to their target-domain models with limited application-specific data using a transfer learning technique. Meanwhile, the federated learning scheme is deployed within a group to further enhance the accuracy of the application-specific model. The integrated framework allows model sharing across the central server and different smart devices without exposing any raw data and, hence, protects the data privacy. Two public datasets, COCO and PETS2009, which represent the source and target applications, are employed for evaluations. The simulation results show that the proposed method outperforms two state-of-the-art machine learning approaches by achieving better learning efficiency and accuracy.

This article presents a new method that employs the active disturbance rejection control (ADRC) paradigm to handle the control problem for a class of nonlinear systems with unknown disturbances and measurement noise. An improved extended state observer (ESO) is devised to separate disturbance estimation from state reconstruction and a new parameter is used to adjust disturbance-rejection performance independently. This increases the flexibility of system design and makes the gain of the improved ESO (IESO) small compared to that of a conventional ESO in the ADRC. A low-pass filter is added to the IESO. This changes the slope of the magnitude curve of the Bode diagram from $-20$ dB/dec to $-40$ dB/dec at high frequencies for measurement noise, which improves the noise-suppression performance. Comparison with a conventional ESO reveals the superiority of the IESO in disturbance rejection and noise suppression. The presented method is validated by simulation and experimental results.

Wireless sensor networks (WSNs) are usually deployed in the target area to carry out the detecting or monitoring tasks. Sweep coverage is an important concern of WSNs that more targets in the area can be monitored with fewer mobile sensor nodes. With the rapid development of unmanned aerial vehicle (UAV) technology, UAV has been increasingly used in military and civil fields. UAVs can be regarded as sensor nodes to perform specific tasks. Due to the limited battery lifetime of UAV in flight mission, it is difficult to achieve full coverage of all targets in a large-scale monitoring scenario. In this paper, we study how to plan the paths of multiple UAVs for sweep coverage. Based on the background of forest fire early warning and monitoring, we consider a Min-Time Max-Coverage (MTMC) issue in sweep coverage where a set of UAVs are dispatched to efficiently patrol the targets in the given area to achieve maximum coverage in minimum time. We establish a mathematical model considering the different coverage quality requirements of the targets in the given area. Then, we analyze the characteristics of the model and propose a heuristic algorithm Weighted Targets Sweep Coverage (WTSC) to find the optimal path, which considers the weights of the targets and the performance constraints of UAVs. Finally, we provide various numerical experiments and comparisons with several previous works to validate the superiority of the proposed algorithm.

Exploiting the possibility of an unmanned aerial vehicle (UAV) as a powerful tool for the Internet of Things applications, such as intelligent agricultural monitoring, intelligent transportation monitoring, etc., has gradually become a hot research topic at home and abroad. While some optimization algorithms have been devised to plan the flight route of UAVs, there are still some problems with the feasibility and effectiveness of these algorithms. This paper presents a solution to the UAV path planning problem for target coverage task in a dynamic environment. The methodology applies a greedy allocation strategy for task assignment and an improved ant colony optimization algorithm based on variable pheromones (ACO-VP) for path planning. First, we specify the optimal number of UAVs for the task and allocate target points to each UAV, through the greedy allocation strategy. Then, to improve the efficiency of path planning, we adjust the pheromone update rule by introducing a variable pheromone enhancement factor and a variable pheromone evaporation coefficient into the ACO algorithm. Moreover, paths are re-planned when the coverage task changes due to the increase of new target points. This method is verified through simulations and compared with other algorithms. The results show that the ACO-VP algorithm is more efficient and effective for UAV path planning than others.

An improved equivalent-input-disturbance (EID) approach is devised to enhance the disturbance-rejection performance for a strictly proper plant with a state delay in a modified repetitive-control system. A gain factor is introduced to construct an improved EID estimator. This increases the flexibility of system design and enables the adjustment of the dynamical performance of disturbance rejection. Moreover, the commutative condition, which is widely used for the conventional EID estimator, is avoided. Thus, it reduces the conservativeness of design by removing the constraints imposed by the commutative condition. The system is divided into two subsystems, and the separation theorem is applied to simplify the design. For one subsystem, the delay information on both the modified repetitive controller and the plant is used to reduce the conservativeness of stability condition. The resulting linear matrix inequality ('MI) is used to find the gain of the state-feedback controller. Another 'MI is derived to design the gains of the state observer and the improved EID estimator for the other subsystem. A case study on a metal-cutting system validates the superiority of the developed method.