This tutorial presents a comprehensive treatment of LAWNs from the joint perspectives of artificial intelligence (AI) and signal processing, and highlights opportunities for LAWNs to evolve into dependable, AI-native infrastructure for the intelligent skies.
Weijie Yuan, Geng Sun, Jia-Cheng Wang et al.· IEEE Journal on Selected Top...· 0 citations
Low-altitude uncrewed aerial vehicle (UAV) communication offers notable advantages over terrestrial base stations in terms of flexibility and deployment efficiency. However, the high likelihood of line-of-sight (LoS) propagation renders the communication links between UAVs and ground users (GUs) particularly susceptible to eavesdropping. To address this issue, we consider an intelligent reflecting surface (IRS)-assisted low-altitude UAV secure communication system, in which communication security is strengthened through adaptive control of the wireless propagation environment, even when eavesdroppers are present. We aim to maximize the secrecy rate of GUs while minimizing the UAV energy consumption by jointly optimizing the continuous UAV trajectory, power allocation, and discrete IRS phase shifts. Considering the dynamic, non-convex, and NP-hard nature of the optimization problem, we propose an agentic artificial intelligence (AI) approach, namely alternating optimization (AO) and generative diffusion model-based deep deterministic policy gradient (AO-GDMDDPG) approach. The proposed agentic AI approach is composed of two cooperative agents that operate over a hybrid and high-dimensional decision space, in which the UAV agent adopts a generative AI (GenAI)-enhanced deep reinforcement learning (DRL) method to optimize continuous decision variables, whereas the IRS agent relies on the AO method to determine discrete IRS phase shifts. Simulation results demonstrate the superiority of the AO-GDMDDPG approach over benchmark algorithms with respect to secrecy rate improvement and UAV energy consumption reduction.
Wenwen Xie, Geng Sun, Jiahui Li et al.· IEEE Transactions on Cogniti...· 0 citations
RecoverFly is proposed, a failure-aware RL post-training framework for end-to-end UAV-VLA policies that adapts token-level RL for stable optimization of grammar-constrained autoregressive UAV actions, revisits unresolved failure cases to strengthen corrective learning and sample utilization, and combines a two-stage long-tail scene curriculum with reference-policy regularization to improve scene adaptation while preserving acquired capabilities.
Boxiong Wang, Hui Kang, Geng Sun et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
