Unmanned Aerial Vehicle (UAV) formations play a vital role in modern mission-critical applications, such as disaster response, surveillance, and Search and Rescue (SAR) operations. Ensuring the operational health of Unmanned Aerial Vehicles (UAVs) in the most often dynamic environments in which they are deployed requires a robust, scalable, and realtime fault detection mechanism to ensure mission integrity. In this paper, we propose an Edge-Enhanced Graph Attention Network (EE-GAT) model for fault detection in a dynamic UAV formation. The proposed method integrates node-level features such as altitude, velocity, battery level, with edge-level network proximity measures (RSSI and SINR) into an attention-based graph learning framework to enable accurate detection of UAV position status in a leader-follower topology. Through extensive simulations in NS3 v3.42, we demonstrate that our EE-GAT achieves a high fault detection accuracy of up to $\mathbf{9 6. 1 \%}$. Our approach outperforms baseline models, offering a lightweight, and cost-effective solution for real-time UAV health monitoring in dynamic and large-scale formations.

The deployment of 5G networks has ushered in a new era for technologies that demand Ultra Reliable Low Latency (URLL) networks. However, the User Equipment (UE) responsible for handling such applications may not always have the computational capacity to meet the demanding requirements of URLL implementations. In such scenarios, certain tasks can be offloaded to a high-performance computing network known as Multi-access Edge Computing (MEC), located near the UE at a base station. Nonetheless, MEC servers themselves have computational limitations, and they can’t handle all the tasks from all the UEs efficiently. To address this challenge, we propose an efficient task offloading approach based on asynchronous Meta Reinforcement Learning (MRL). Our method involves developing an algorithm that prioritizes and organizes UE tasks based on their importance and interdependencies, and a Deep asynchronous MRL algorithm models the task offloading policy. What sets our model apart is its ability to adapt quickly to a variety of environments, whether they are heterogeneous or homogeneous. Our simulation results show that our solution has a 60% overall performance improvement in terms of training latency, energy consumption, and memory and CPU time usage compared to a benchmark.

There is a definite possibility that, in a recent future, Unmanned Aerial Vehicles (UAVs) will form the backbone of any smart city in terms of automation and networking. One approach to extend the UAVs’ resources spectrum is to provide a mean for them to opportunistically recharge and connect to otherwise unreachable networks: provide Training and Recharge Areas (TRAs). In these dedicated areas, the UAVs could dock to Energy and Data Dispensers (EDD) devices to resupply their batteries and exploit a high-speed connection. To autonomously move through the smart city while accomplishing a set of given tasks but, at the same time, consider visiting the EDDs, is part of a tridimensional trajectory planning problem that needs to be addressed. In this paper, we formally define the combinatorial optimization problem representing the trajectory planning. We consider the case in which more than one UAV can be connected with the same EDD at the same time, by properly addressing the assignment of the bandwidth. Through simulative investigation, realistic values for the solution of the optimization problem are found. The behavior of the proposed model is compared with an ”online” approach that does not require the same resources and knowledge and whose evaluation and comparison with the ”offline” approach are performed through network simulation.

In the context of wireless networked robotics, complex missions, such as autonomous Search and Rescue, may require the use of multiple Unmanned Aerial Vehicles (UAVs) to achieve higher efficiency and Quality of Service (QoS) while assuring mission-specific imperatives like the maximization of area covered in a single passage of the fleet over area subsections. In this paper, we present a learning-based formation control protocol that adapts the principle of Q-learning to pilot an autonomous fleet of networked UAVs to maintain formation throughout a mission where large quantities of data need to be exchanged. Also, the protocol tries to ensure rotational formation control by leveraging only the signal strength extrapolated from the UAV communications. A leader-follower model is used to control the fleet. One UAV serves as the leader, and the remaining as the followers. The followers use the Received Signal Strength Indicator (RSSI) values obtained from their neighbors to autonomously determine the leader's direction of movement and maintain formation orientation to avoid area coverage overlapping. We carried out several simulation experiments to evaluate the performance of the proposed scheme in terms of QoS and convergence time of the formation under varying velocities.

This paper formulates the Sport Event Filming with Connectivity Constraints (SEF-C2) problem. The SEF-C2problem is an event coverage problem, which exploits a team of Unmanned Aerial Vehicles (UAVs) over a limited field in order to track the movements of an object (e.g., of the ball) and to deliver a video stream of the events (e.g., ball passes, goals) to the spectators meeting certain timeliness and video quality criteria. Assuming a priori knowledge of the whole sequence of actions, first a novel mathematical model is introduced that determines a sequence of movements for the UAVs, such that the timeliness of the filming is maximized and the total traveled distance is minimized. Then, dynamic, artificial potential function based, distributed UAV movement schemes that have no a priori knowledge of the sequence of game actions are proposed to optimize networking performance. Extensive simulations are used to analyze the performance in terms of video transmission quality and show that the proposed schemes outperform existing schemes.

In this paper, an innovative approach based on a Q-learning algorithm to allow a single operator to control a fleet of Unmanned Aerial Vehicles (UAVs) is presented. To follow an independently-controlled UAV (the leader), all the other UAVs (the followers) use only the radio signal strength values received during wireless ad-hoc communications among themselves, without the need for any infrastructure, any localization technique or any additional device. By applying the proposed behavior-based control scheme in real-time, the fleet formation can be perfectly maintained. The solution has been implemented through the definition of a new protocol and has been tested using ns-3. Experiments highlight the efficiency and effectiveness of the proposed method.

The merging of networking and control fields has always brought interesting innovations but the tools and structures for proper and easy management of experiments still lag behind. Different solutions have been proposed to handle general control problems and, more in detail, for fine control of UAVs (Unmanned Aerial Vehicles) dynamics. They lack, however, an efficient and detailed network-side simulation, usually available only on dedicated software. On the other hand, current advancements in network simulations suites often do not include an accurate simulation of controlled systems. In the middle 2010s, integrated solutions are still lacking. For these reasons, in this paper we propose a simulation architecture for networked control systems. The architecture is based on well-known solutions in both the fields of networking simulation and UAV control simulation. We integrate them into a compact and efficient solution that shows scalability features and negligible architectural delays, as experimental results demonstrate.

A software suite that is capable to simulate Unmanned Aerial Vehicle (UAV) dynamics and, at the same time, network-side behaviors is still lacking in literature. Despite the recent merge between the fields of networking and control, the existing solutions are either dedicated to only one between the aforementioned topics. In this demonstration paper, we describe a novel simulation architecture to implement distributed networked control systems. Our proposal consists in the integration of already-present and validated solutions into a compact package that features scalability and negligible architectural delays.

Due to their deployment flexibility, Unmanned Aerial Vehicles have been found suitable for many application areas, one of them being air pollution monitoring. In fact, deploying a fleet of Unmanned Aerial Vehicles (UAVs) and using them to take environmental samples is an approach that has the potential to become one of the key enabling technologies to enforce pollution control in industrial or rural areas. In this paper, we propose to use an algorithm called Pollution-driven UAV Control (PdUC) that is based on a chemotaxis metaheuristic and a Particle Swarm Optimization (PSO) scheme that only uses local information. Our approach will be used by a monitoring Unmanned Aerial Vehicle to swiftly cover an area and map the distribution of its aerial pollution. We show that, when using PdUC, an implicit priority is applied in the construction of pollution maps, by focusing on areas where the pollutants' concentration is higher. In this way, accurate maps can be constructed in a faster manner when compared to other strategies. We compare PdUC against various standard mobility models through simulation, showing that our protocol achieves better performances, by finding the most polluted areas with more accuracy, within the time bounds defined by the UAV flight time.

Leveraging the concept of controlled node mobility in this paper we develop an algorithm for tracking and controlling proximity malware propagation in Wireless Sensor Networks (WSN). Our proposal aims at: (i) notifying the nodes of malwarepropagation, (ii) leading a flying robot along a path in the WSNthat guarantees the minimum recovery time to (iii) heal the infected nodes. We formulate the targeted curing problem as a binary integer problem and determine the optimal solution by a central solver. We use the analytical result as a benchmark to evaluate the recovery time of the proposed solution. The achieved results show a satisfactory performance in terms of tracking the presence of an ongoing epidemic and healing the nodes.