Adaptable Deep Joint Source-and-Channel Coding for Small Satellite Applications

TL;DR

The paper tackles bandwidth-limited downlinks for small satellites by developing AdaJscc, an attention-augmented, single-network deep joint source-channel coding scheme trained with a realistic Fontán-based satellite channel model and Sentinel-2 data. By replacing multiple channel-specific networks with a parametrizable architecture, it uses attention modules to adapt to diverse channel states while maintaining compression via $k/n$. Empirical results show AdaJscc achieves PSNR comparable to Baseline methods that rely on separate networks, while substantially reducing storage and offering robustness under channel-state mismatch, particularly at lower elevation angles. This work advances practical onboard JSCC for low-Earth-orbit missions by enabling adaptive, efficient communication under realistic, dynamic channel conditions.

Abstract

Earth observation with small satellites serves a wide range of relevant applications. However, significant advances in sensor technology (e.g., higher resolution, multiple spectrums beyond visible light) in combination with challenging channel characteristics lead to a communication bottleneck when transmitting the collected data to Earth. Recently, joint source coding, channel coding, and modulation based on neuronal networks has been proposed to combine image compression and communication. Though this approach achieves promising results when applied to standard terrestrial channel models, it remains an open question whether it is suitable for the more complicated and quickly varying satellite communication channel. In this paper, we consider a detailed satellite channel model accounting for different shadowing conditions and train an encoder-decoder architecture with realistic Sentinel-2 satellite imagery. In addition, to reduce the overhead associated with applying multiple neural networks for various channel states, we leverage attention modules and train a single adaptable neural network that covers a wide range of different channel conditions. Our evaluation results show that the proposed approach achieves similar performance when compared to less space-efficient schemes that utilize separate neuronal networks for differing channel conditions.

Figures (8)

• Figure 1: Example images from the Sentinel-2 mission. (Credit: processed by ESA, CC BY-SA 3.0 IGO)
• Figure 2: Overview of our communication architecture and comparison with traditional separate-encoder designs.
• Figure 3: Network architecture overview.
• Figure 4: Specific block architectures used within the encoder-decoder network.
• Figure 5: psnr achieved by the Baseline for different environments, states, and compression ratios with 40° elevation angle.