Edge Intelligence and Multi-Agent Learning for Connected, Autonomous and Aerial Networks

Connected networks now carry a growing share of intelligent decision-making. Smart cities, industrial IoT and connected vehicles generate data continuously and need answers within milliseconds, often with limited energy, bandwidth and unreliable links. Aerial networks add a further layer. They comprise airborne platforms operating within Earth’s atmosphere, such as unmanned aerial vehicles (UAVs), drones, balloons and high-altitude platform stations (HAPS). These platforms can act as mobile sensors, relays and edge nodes, extending coverage where terrestrial infrastructure is missing, damaged or overloaded. They also face tight onboard energy, changing topology and intermittent connectivity.Sending everything to the cloud no longer meets these latency, privacy and cost requirements. Edge intelligence moves learning and inference closer to the data source, while multi-agent learning, including Multi-Agent Deep Reinforcement Learning (MADRL), lets many devices and airborne platforms coordinate in changing and uncertain conditions. In parallel, the Social Internet of Things (SIoT) provides ways for devices to build trust, share context and recommend services to one another.Several questions remain open as these technologies converge. How should computation be split between device, edge, airborne platform and cloud under strict time and energy limits? How can cooperating edge agents learn reliably when nodes join, leave or fail, or when some participants are unreliable or malicious? How can devices decide whom to trust and which services to use? This Research Topic focuses on cooperation between edge devices and agents, rather than on the operation of core or backbone network infrastructure. Our goal is to create a forum for new methods, rigorous evaluations and practical systems that make edge-based, cooperative and trustworthy intelligence work at scale in connected, autonomous and aerial networks.We welcome Original Research, Review, Systematic Review, Methods, Perspective, Mini Review and Technology and Code articles that address these challenges. Topics of interest include, but are not limited to:- Task offloading and edge-cloud orchestration under latency, energy and accuracy constraints- Selective inference offloading and resource allocation for real-time IoT sensing- Fog and edge node placement and deployment, including UAV-assisted and mobile edge networks- Aerial edge computing with UAVs, drones, balloons and HAPS as mobile compute and relay nodes- Multi-agent deep reinforcement learning for cooperative scheduling, offloading and resource management among edge devices and airborne platforms- Federated and privacy-preserving learning with heterogeneous, unreliable or adversarial clients- Metaheuristic, optimization and learning-based approaches to edge network design- Trust management, secure collaboration and service recommendation in the Social Internet of Things- Vehicular networks and the Internet of Vehicles, including V2V content dissemination and cooperative sensing- Swarm coordination, trajectory planning and communication in UAV and other autonomous aerial fleets- Energy-aware offloading and scheduling on resource-constrained edge and airborne devices- Robustness of multi-agent and federated learning to poisoning attacks, faulty nodes and adversarial participants- Digital twins, simulators and benchmarks for cooperative edge AI, including aerial scenarios Topic editors: Amar Khelloufi, Abdenacer Naouri Source: https://www.frontiersin.org/research-topics/87053/edge-intelligence-and-multi-agent-learning-for-connected-autonomous-and-aerial-networks