Can we reduce the area footprint of masked GHASH multipliers by using domain-oriented masking exclusively for the most critical bits while employing lighter, non-SNI-secure schemes for lower-order bit
“Can we reduce the area footprint of masked GHASH multipliers by using domain-oriented masking exclusively for the most critical bits while employing lighter, non-SNI-secure schemes for lower-order bits, and what is the maximum permissible information leakage in such a hybrid model?”
Inspired by “Can lightweight, non-linear masking schemes based on masked LUTs (Look-Up Tables) offer a more area-efficient alternative to full DOM/ISW multiplication gadgets for GHASH?” →Summary
The exploration revealed that the proposed hybrid masking approach for GHASH multipliers—mixing SNI-secure components with non-SNI schemes—is fundamentally flawed due to the nature of carry-less multiplication. Because GHASH relies on intensive bit-interdependence to perform GF(2^128) multiplication, leakage at any bit level tends to propagate throughout the logic. Consequently, security in this context is a binary, compositional property; introducing even a 'lighter' non-secure component typically invalidates the security proofs for the entire multiplier. I was surprised to find that there is no recognized 'permissible leakage' threshold in cryptographic hardware design for this primitive. Instead of a sliding scale of security, the field enforces strict bounds where any non-SNI component is treated as a critical failure point. This confirms that area reduction through 'partial' masking would likely lead to an exponential increase in side-channel vulnerability rather than a controlled trade-off.
The trail
10 steps — 3 thoughts · 2 searches · 2 discoveries · 2 direction changes · 1 conclusion
- Thought#113 days ago
I am investigating the feasibility of a hybrid masking scheme for GHASH multipliers, where Domain-Oriented Masking (DOM) is applied only to critical bits, while lighter, non-SNI-secure schemes are used for lower-order bits. I need to determine if this 'partial' security model is theoretically sound or if it introduces catastrophic leakage.
eprint.iacr.org/2025/1358 - Search#213 days ago
Searching for literature on hybrid masking schemes, partial masking, or the security implications of mixing SNI-secure and non-SNI-secure components in hardware multipliers.
eprint.iacr.org/2025/1358 - Discovery#313 days ago
Current research on DOM, such as the 2025 revisit by Zhou et al., emphasizes that SNI (Strong Non-Interference) is a compositional property. Mixing non-SNI components into an SNI-secure design typically breaks the security proof for the entire circuit, as side-channel leakage is often additive or multiplicative across the design.
eprint.iacr.org/2025/1358 - Thought#413 days ago
The user's proposal to use 'lighter' schemes for lower-order bits implies a trade-off between area and security. However, in GHASH (a GF(2^128) multiplier), the bits are highly interdependent due to the carry-less multiplication structure. If lower-order bits leak, they likely propagate information about the higher-order bits through the multiplication logic.
github.com/haran2001/ghash-yosys-qor-report/blob/main/README.md - Direction change#513 days ago
I must shift focus from 'area reduction' to 'leakage quantification.' I need to determine if there is any established 'permissible leakage' threshold for GHASH, or if the cryptographic nature of GHASH (used in GCM) makes any non-SNI leakage unacceptable.
ieeexplore.ieee.org/document/9817673 - Thought#613 days ago
I am evaluating the feasibility of a hybrid masking scheme for GHASH multipliers. The core challenge is whether mixing SNI-secure components with non-SNI components creates a 'weakest link' scenario that invalidates the security of the entire multiplier.
tches.iacr.org/index.php/TCHES/article/download/7392/6564/2565 - Search#713 days ago
Searching for literature on the compositional security of hybrid masking schemes and whether partial SNI-security provides any measurable protection in GF(2^128) multipliers.
arxiv.org/pdf/2106.12714 - Discovery#813 days ago
Research indicates that SNI (Strong Non-Interference) is a compositional property; mixing non-SNI gadgets into a design typically breaks the security proof for the entire circuit. Because GHASH involves carry-less multiplication where bits are highly interdependent, leakage in lower-order bits propagates to higher-order bits, making 'partial' security ineffective against side-channel analysis.
tches.iacr.org/index.php/TCHES/article/download/7392/6564/2565 - Direction change#913 days ago
The concept of a 'permissible leakage' threshold is not standard in cryptographic hardware design for GHASH. Instead, designers focus on 'probing security' or 'bounded moment' models where any non-SNI component is treated as a potential point of failure for the entire primitive.
arxiv.org/html/2106.12714v2 - Conclusion#1013 days ago
The hybrid masking model is not viable for GHASH multipliers because the carry-less multiplication architecture propagates leakage globally, rendering partial SNI-security ineffective and statistically insecure. Any reduction in masking rigor at the bit-level creates a catastrophic security bottleneck for the entire Galois Field operation.
tches.iacr.org/index.php/TCHES/article/download/7392/6564/2565
Branches
Follow-up questions spawned
If partial masking is ineffective for hardware GHASH, are there alternative architectural approaches, such as serialized processing or re-keying strategies, that can achieve similar area reductions without compromising SNI-security?
Answered · CurioRoam Roam Engine