Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Robust Semantic Transmission for Low-Altitude UAVs: Predictive Channel-Aware Scheduling and Generative Reconstruction

Jijia Tian, Junting Chen, Pooi-Yuen Kam

TL;DR

The paper tackles robust semantic downlink for low-altitude UAVs under bandwidth constraints and highly time-varying A2G channels. It introduces a predictive semantic framework that decouples content into a deterministic structural component and a stochastic texture component using a Structure-Texture Variational Autoencoder (ST-VAE), along with a channel-aware scheduler that prioritizes structure transmission and employs a receiver-side conditional generative prior for texture synthesis when blocks are missing. The key contributions include (i) a trajectory-informed SNR predictor feeding a per-slot budgeting and block-scheduling algorithm, (ii) explicit semantic encoding with a structure-first policy and texture completion via a conditional prior, and (iii) substantial performance gains (up to 5.6 dB PSNR) and robustness to prediction mismatch in simulations using COCO data and a realistic A2G channel model. Collectively, the approach mitigates cliff effects inherent in coupled DeepJSCC, improving perceptual and structural fidelity in challenging UAV downlink scenarios with strict bandwidth budgets.

Abstract

Unmanned aerial vehicle (UAV) downlink transmission facilitates critical time-sensitive visual applications but is fundamentally constrained by bandwidth scarcity and dynamic channel impairments. The rapid fluctuation of the air-to-ground (A2G) link creates a regime where reliable transmission slots are intermittent and future channel quality can only be predicted with uncertainty. Conventional deep joint source-channel coding (DeepJSCC) methods transmit coupled feature streams, causing global reconstruction failure when specific time slots experience deep fading. Decoupling semantic content into a deterministic structure component and a stochastic texture component enables differentiated error protection strategies aligned with channel reliability. A predictive transmission framework is developed that utilizes a split-stream variational codec and a channel-aware scheduler to prioritize the delivery of structural layout over reliable slots. Experimental evaluations indicate that this approach achieves a 5.6 dB gain in peak signal-to-noise (SNR) ratio over single-stream baselines and maintains structural fidelity under significant prediction mismatch.

Robust Semantic Transmission for Low-Altitude UAVs: Predictive Channel-Aware Scheduling and Generative Reconstruction

TL;DR

The paper tackles robust semantic downlink for low-altitude UAVs under bandwidth constraints and highly time-varying A2G channels. It introduces a predictive semantic framework that decouples content into a deterministic structural component and a stochastic texture component using a Structure-Texture Variational Autoencoder (ST-VAE), along with a channel-aware scheduler that prioritizes structure transmission and employs a receiver-side conditional generative prior for texture synthesis when blocks are missing. The key contributions include (i) a trajectory-informed SNR predictor feeding a per-slot budgeting and block-scheduling algorithm, (ii) explicit semantic encoding with a structure-first policy and texture completion via a conditional prior, and (iii) substantial performance gains (up to 5.6 dB PSNR) and robustness to prediction mismatch in simulations using COCO data and a realistic A2G channel model. Collectively, the approach mitigates cliff effects inherent in coupled DeepJSCC, improving perceptual and structural fidelity in challenging UAV downlink scenarios with strict bandwidth budgets.

Abstract

Unmanned aerial vehicle (UAV) downlink transmission facilitates critical time-sensitive visual applications but is fundamentally constrained by bandwidth scarcity and dynamic channel impairments. The rapid fluctuation of the air-to-ground (A2G) link creates a regime where reliable transmission slots are intermittent and future channel quality can only be predicted with uncertainty. Conventional deep joint source-channel coding (DeepJSCC) methods transmit coupled feature streams, causing global reconstruction failure when specific time slots experience deep fading. Decoupling semantic content into a deterministic structure component and a stochastic texture component enables differentiated error protection strategies aligned with channel reliability. A predictive transmission framework is developed that utilizes a split-stream variational codec and a channel-aware scheduler to prioritize the delivery of structural layout over reliable slots. Experimental evaluations indicate that this approach achieves a 5.6 dB gain in peak signal-to-noise (SNR) ratio over single-stream baselines and maintains structural fidelity under significant prediction mismatch.
Paper Structure (19 sections, 29 equations, 6 figures)

This paper contains 19 sections, 29 equations, 6 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Geometric Configuration and Trajectory
  4. Probabilistic A2G Channel Model
  5. LOS/NLOS State
  6. Path Loss and Shadowing
  7. Signal Transmission Model
  8. Data-Driven SNR Prediction
  9. Semantic Rate Distortion Perception Formulation
  10. Synonymity Based Reconstruction Model
  11. Abstract Objective Under Channel Use Budgets
  12. Neural Realization ST-VAE and Predictive Scheduling
  13. ST-VAE Semantic Codec
  14. Predictive Scheduler Under Per Slot Sample Budgets
  15. Simulation Results
  16. ...and 4 more sections

Figures (6)

  • Figure 1: Illustration of the time-varying UAV downlink induced by geometry and blockage. (a) The UAV trajectory and an obstacle lead to LOS/NLOS transitions. (b) The resulting time-slotted SNR sequence exhibits an outage interval where $\gamma_k < \gamma_{\min}$ (no transmission allowed) and a usable-slot set $\mathcal{K}_{\text{use}}$ for scheduling based on SNR prediction.
  • Figure 2: Overview of the ST-VAE framework.
  • Figure 3: Peak signal to noise ratio versus average realized signal to noise ratio.
  • Figure 4: Structural similarity index versus average realized signal to noise ratio.
  • Figure 5: Peak signal to noise ratio versus prediction mismatch standard deviation in decibel domain.
  • ...and 1 more figures