Information and Communications Security

Ding Wang, Moti Yung, Zheli Liu, Xiaofeng ChenLecture Notes in Computer Science 2023 · DOI · OpenAlex

Introduction

The 25th International Conference on Information and Communications Security (ICICS 2023) represents a significant milestone in the field of cybersecurity. These proceedings offer a comprehensive snapshot of contemporary research, spanning from foundational cryptanalysis to the innovative intersection of artificial intelligence and privacy.

Springer, the publisher of the ICICS 2023 proceedings.

The Challenge of Data in Use

As digital infrastructure expands into the realms of the Internet of Things (IoT), Virtual Reality (VR), and decentralized finance, traditional security models face mounting pressure. A central theme of this volume is the "protection of data in use." While robust solutions exist for data in transit (TLS) and data at rest (BitLocker), securing data during active processing remains a significant hurdle.

Advanced techniques like Fully Homomorphic Encryption (FHE) are often computationally expensive, while hardware-based Trusted Execution Environments (TEEs) offer speed but present a larger attack surface. The research presented here aims to bridge these gaps, making high-level security practical for high-performance, real-world applications.

Methodological Approaches

The proceedings categorize research into specialized domains using both "breaking" (cryptanalysis) and "building" (system design) methodologies:

  • Cryptography: Authors employ SAT-aided differential cryptanalysis for lightweight block ciphers and explore new attacks on post-quantum candidates like SIDH.
  • Machine Learning for Security: Techniques like LightGBM are used for cipher classification, while deep learning is applied to detect deepfakes and identify illegal accounts on the Ethereum blockchain.
  • Applied Systems: The design of "oblivious" in-enclave processing systems, such as TimeClave, which hides access patterns from host systems.
  • AI Security: Research into adversarial perturbations, backdoor attacks on neural networks, and "graph unlearning" to remove sensitive data from trained models.

Key Insights

  • The TEE-Crypto Hybrid: A recurring theme is the synergy between hardware and software security. TEEs provide the necessary speed for processing, while cryptographic proofs offer verifiable guarantees that the hardware has not been compromised.
  • AI as a Dual-Use Tool: AI is increasingly used for both sophisticated attacks (e.g., inferring passwords through acoustic side channels) and defense (e.g., detecting deepfakes via adaptive attention).
  • Side-Channel Persistence: Despite stronger encryption, contextual vulnerabilities like energy consumption, timing, and acoustics remain significant threats for IoT and mobile devices, allowing adversaries to infer sensitive actions without breaking the encryption itself.

Conference Outcomes and Organization

ICICS 2023 maintained a rigorous selection process, accepting only 38 full papers and 6 short papers from 210 submissions, resulting in a competitive 20.95% acceptance rate.

Organizing committee of ICICS 2023, including Program and General Chairs.

Notable awards included:

  • Best Paper: "BDTS: Blockchain-based Data Trading System," which proposes a secure framework for the burgeoning data economy.
  • Best Student Paper: "An Efficient Attack on Dimension Two SIDH," providing critical feedback on post-quantum cryptographic schemes.

Conclusion and Future Directions

While the proposed solutions show promise, scaling them for global deployment remains a challenge. Future work suggested by the proceedings includes developing more robust threat models for TEEs, improving the efficiency of privacy-preserving AI, and addressing cross-domain blockchain challenges such as quantifying maximal extractable value (MEV).

Summary updated

BibTeX
@inproceedings{wang2023information,
  author = {Ding Wang and Moti Yung and Zheli Liu and Xiaofeng Chen},
  title = {{Information and Communications Security}},
  booktitle = {Lecture Notes in Computer Science},
  year = {2023},
  doi = {10.1007/978-981-99-7356-9},
}
Figure