|
PQ-WB-KEM: White-Box ML-KEM-768 1.0
PQ-WB-KEM: Toward a White-Box Construction of ML-KEM-768 with Arithmetic Masking for M2M Communications - Mathematics (MDPI)
|
Measure the stored size of the two obfuscation layers wrapping the base multiply – the coefficient shuffle and the runtime masking. More...
Go to the source code of this file.
Namespaces | |
| namespace | obfuscation_storage |
Functions | |
| obfuscation_storage.build_shuffle (rng) | |
| One coefficient shuffle pi and its inverse, as stored uint16 tables. | |
| obfuscation_storage.main () | |
| Measure the obfuscation-layer storage and write the summary JSON. | |
| obfuscation_storage.read_core () | |
| Read the measured core footprints from concrete_security_bounds.json. | |
| Path | obfuscation_storage.results_dir () |
| Repository results directory. | |
Variables | |
| int | obfuscation_storage.BYTES_PER_ENTRY = 2 |
| int | obfuscation_storage.K = 3 |
| int | obfuscation_storage.MB = 1_000_000 |
| int | obfuscation_storage.N = 256 |
| int | obfuscation_storage.SEED = 42 |
Measure the stored size of the two obfuscation layers wrapping the base multiply – the coefficient shuffle and the runtime masking.
The paper's storage breakdown previously carried the shuffle and masking cost as a bare estimate. This produces it. The shuffle is a permutation of the N=256 NTT coefficients of the secret (built by np.random.permutation(N) in algebraic_attack_analysis.py; the manuscript calls it a coefficient shuffle), applied per secret polynomial (k=3 for ML-KEM-768). It is stored with its inverse at the uint16 width the base and encoding tables use. Masking stores nothing: the three arithmetic shares are drawn at runtime per decapsulation (whitebox_reference.py three_share), so its stored size is zero. The deployment footprint is therefore the measured core (concrete_security_bounds.json) plus a few kilobytes.
Definition in file obfuscation_storage.py.