Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Strategic Application of AIGC for UAV Trajectory Design: A Channel Knowledge Map Approach

Chiya Zhang, Ting Wang, Rubing Han, Yuanxiang Gong

TL;DR

The paper tackles the challenge of accurate channel loss prediction and efficient UAV trajectory design under data scarcity in UAV-IoT wireless networks. It introduces an AIGC-driven Channel Knowledge Map (CKM) that leverages a Wasserstein GAN for simulation data augmentation and a knowledge-driven CKM architecture with LoS-based physics and an environment encoder to improve channel gain predictions. A reinforcement-learning framework (MDP/PPO) uses CKM-derived channel gains to optimize UAV trajectories while satisfying power, motion, and QoS constraints, with the CKM guiding resource allocation and decision-making. Experimental results in synthetic urban environments show that AIGC-enhanced CKM predictions reduce channel gain uncertainty and enable more effective trajectory planning, outperforming traditional LoS-based and BCD-based baselines in terms of throughput and robustness. Overall, the approach demonstrates how synthetic data and physics-informed learning can substantially improve UAV-assisted wireless performance while mitigating data collection costs, with potential impact on 6G-era aerial networks.

Abstract

Unmanned Aerial Vehicles (UAVs) are increasingly utilized in wireless communication, yet accurate channel loss prediction remains a significant challenge, limiting resource optimization performance. To address this issue, this paper leverages Artificial Intelligence Generated Content (AIGC) for the efficient construction of Channel Knowledge Maps (CKM) and UAV trajectory design. Given the time-consuming nature of channel data collection, AI techniques are employed in a Wasserstein Generative Adversarial Network (WGAN) to extract environmental features and augment the data. Experiment results demonstrate the effectiveness of the proposed framework in improving CKM construction accuracy. Moreover, integrating CKM into UAV trajectory planning reduces channel gain uncertainty, demonstrating its potential to enhance wireless communication efficiency.

Strategic Application of AIGC for UAV Trajectory Design: A Channel Knowledge Map Approach

TL;DR

The paper tackles the challenge of accurate channel loss prediction and efficient UAV trajectory design under data scarcity in UAV-IoT wireless networks. It introduces an AIGC-driven Channel Knowledge Map (CKM) that leverages a Wasserstein GAN for simulation data augmentation and a knowledge-driven CKM architecture with LoS-based physics and an environment encoder to improve channel gain predictions. A reinforcement-learning framework (MDP/PPO) uses CKM-derived channel gains to optimize UAV trajectories while satisfying power, motion, and QoS constraints, with the CKM guiding resource allocation and decision-making. Experimental results in synthetic urban environments show that AIGC-enhanced CKM predictions reduce channel gain uncertainty and enable more effective trajectory planning, outperforming traditional LoS-based and BCD-based baselines in terms of throughput and robustness. Overall, the approach demonstrates how synthetic data and physics-informed learning can substantially improve UAV-assisted wireless performance while mitigating data collection costs, with potential impact on 6G-era aerial networks.

Abstract

Unmanned Aerial Vehicles (UAVs) are increasingly utilized in wireless communication, yet accurate channel loss prediction remains a significant challenge, limiting resource optimization performance. To address this issue, this paper leverages Artificial Intelligence Generated Content (AIGC) for the efficient construction of Channel Knowledge Maps (CKM) and UAV trajectory design. Given the time-consuming nature of channel data collection, AI techniques are employed in a Wasserstein Generative Adversarial Network (WGAN) to extract environmental features and augment the data. Experiment results demonstrate the effectiveness of the proposed framework in improving CKM construction accuracy. Moreover, integrating CKM into UAV trajectory planning reduces channel gain uncertainty, demonstrating its potential to enhance wireless communication efficiency.

Paper Structure

This paper contains 12 sections, 19 equations, 5 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. The UAV-assisted Communication Framework
  4. Channel Model
  5. Formulated Problem
  6. AIGC-Enhanced Environment-Aware UAV Trajectory Design
  7. AIGC for Simulation Data Augmentation
  8. AIGC for Channel Gain Prediction
  9. MDP Formulation
  10. AIGC for UAV Trajectory and Communication Design
  11. Simulation Results
  12. Conclusion

Figures (5)

  • Figure 1: AIGC applications for communication design of the UAV-assisted system.
  • Figure 2: Validation of AIGC applications in UAV trajectory design.
  • Figure 3: Distribution visualization of original and augmented data, with 'xG, yG, zG' for GUs' positions, 'xU, yU, zU' for GUs' positions, 'd' for distance between them and 'g' for the channel gains. The pink boxed section contains distribution plots for each variable as well as the relationship plots between distance and gain.
  • Figure 4: Radar plot of five parameters to assess the networks: training time, inference time, MAPE, MSE Reduction, model parameters(lower values indicate better performance for all metrics except MSE reduction).
  • Figure 5: Generated Trajectories Comparison of Four methods. Los-PPO: PPO algorithm with LoS model; KDCKM-PPO: PPO algorithm with KD-CKM; los-BCD: BCD algorithm with LoS model; los-BCD-loose: los-BCD with start point not fixed.