Context-Aware Information Transfer via Digital Semantic Communication in UAV-Based Networks

TL;DR

This work tackles bandwidth-limited UAV-based surveillance by introducing DSC-UAV, a context-aware digital semantic communication framework that uses a prompt-guided ViT encoder and FiLM-conditioned CNN decoder to transmit task-relevant visual semantics. It couples this semantic layer with dynamic, multi-UAV relaying and a Truncated Quantile Critic (TQC) reinforcement learning agent to jointly optimize UAV trajectories, data compression, and relay allocations, aiming to minimize average AoI and maximize the minimum Semantic Structural Similarity (SSS) across users. Key innovations include soft-to-hard digital quantization for robust digital transmission, OFDM-based ground-to-air and air-to-ground relaying, and a multi-agent MDP formulation that directly ties semantic fidelity to network decisions. Results show up to about 14% AoI reduction and 22% SSS improvement over baselines, demonstrating the practical value of context-driven semantic transmission for resilient, bandwidth-constrained UAV surveillance applications.

Abstract

In smart cities, bandwidth-constrained Unmanned Aerial Vehicles (UAVs) often fail to relay mission-critical data in time, compromising real-time decision-making. This highlights the need for faster and more efficient transmission of only the most relevant information. To address this, we propose DSC-UAV model, leveraging a context-adaptive Digital Semantic Communication (DSC) framework. This model redefines aerial data transmission through three core components: prompt-aware encoding, dynamic UAV-enabled relaying, and user mobility-optimized reinforcement learning. Ground users transmit context-driven visual content. Images are encoded via Vision Transformer combined with a prompt-text encoder to generate semantic features based on the desired context (generic or object-specific). These features are then quantized and transmitted over a UAV network that dynamically relays the data. Joint trajectory and resource allocation are optimized using Truncated Quantile Critic (TQC)-aided reinforcement learning technique, which offers greater stability and precision over standard SAC and TD3 due to its resistance to overestimation bias. Simulations demonstrate significant performance improvement, up to 22% gain in semantic-structural similarity and 14% reduction in Age of Information (AoI) compared to digital and prior UAV-semantic communication baselines. By integrating mobility control with context-driven visual abstraction, DSC-UAV advances resilient, information-centric surveillance for next-generation UAV networks in bandwidth-constrained environments.

Figures (6)

• Figure 1: System Model
• Figure 2: $k^{th}$ Time Slot for $n^{th}$ UAV and $m^{th}$ GU
• Figure 3: Encoder - Decoder Architecture
• Figure 4: Effect of time slot duration $\tau$ on system performance
• Figure 5: Effect of semantic feature length and modulation order