The field of secure communication and computation is rapidly advancing with innovative solutions to longstanding problems. Recent developments have focused on enhancing the scalability and practical security of large-scale quantum key distribution networks, accelerating graph neural networks, and improving the efficiency of fully homomorphic encryption. Notably, researchers have proposed zero-trust relay architectures, adaptive edge sampling strategies, and hybrid CPU-GPU systems to address the limitations of existing methods. These advancements have significant implications for the integration of secure communication and computation into real-world systems. Noteworthy papers include:
- A zero-trust relay design that applies fully homomorphic encryption to perform intermediate OTP re-encryption without exposing plaintext keys.
- AES-SpMM, an adaptive edge sampling SpMM kernel that balances accuracy and speed in graph neural networks.
- PilotANN, a hybrid CPU-GPU system for graph-based approximate nearest neighbor search that achieves significant speedup and memory efficiency.
- CAT, a GPU-accelerated fully homomorphic encryption framework that surpasses existing solutions in functionality and efficiency.
- PP-SND, a novel Privacy-Preserving Secure Neighbor Discovery protocol that enables devices to perform secure neighbor discovery without revealing their actual identities and locations.