The recent advancements in the field of satellite-based AI and IoT security have seen significant innovations, particularly in addressing the unique challenges posed by Low Earth Orbit (LEO) satellite networks and resource-constrained IoT devices. A notable trend is the integration of blockchain technology and federated learning to enhance security and efficiency in satellite communications. This approach leverages decentralized frameworks to manage data heterogeneity and improve model accuracy, while also optimizing energy consumption and system robustness against cyberattacks. Additionally, there is a growing focus on hardware acceleration for zero-knowledge proofs, enabling more efficient cryptographic operations on edge devices. These developments are crucial for future-proofing IoT systems, ensuring secure and efficient cross-domain data sharing, and enhancing the adaptability of authentication protocols in cloud-edge-device collaborative environments.

Noteworthy papers include:

• A sharded blockchain-based federated learning framework for LEO satellite networks that significantly enhances system robustness against attacks.
• A hardware-optimized framework for ZK-friendly hashing on edge devices, outperforming standard CPU implementations by more than 13 times.
• A novel decentralized personalized federated learning framework for heterogeneous LEO satellite constellations, demonstrating superior performance in on-orbit training.