Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys¶
Year: 2025
type: technique collection: schemes status: auto
Links: ePrint 2025/1717
Summary¶
Auto-extracted summary (unreviewed)
The paper proposes encoding a function's look-up table in a vector of polynomials rather than a single test polynomial, representing the additive group Z_q by monic monomial permutation matrices that unify the Alperin-Sheriff-Peikert permutation-matrix and FHEW/TFHE monomial representations, and derives from this a new functional bootstrapping algorithm (BootMMPM) for large plaintexts with polynomially smaller bootstrapping and key-switching keys. Implemented in PALISADE, BootMMPM bootstraps 7-bit plaintexts about 3x faster than TFHE on a desktop (2.6 s vs 10.1 s) and keeps key sizes constant (512 MB + 4.69 GB) while TFHE's grow to 32 GB + 300 GB at 11-bit plaintexts.
Abstract¶
Functional bootstrapping is a core technique in Fully Homomorphic Encryption(FHE). For large plaintext, to evaluate a general function homomorphically over a ciphertext, in the FHEW/TFHE approach, since the function in look-up table form is encoded in the coefficients of a test polynomial, the degree of the polynomial must be high enough to hold the entire table. This increases the bootstrapping time complexity and memory cost, as the size of bootstrapping keys and keyswitching keys need to be large accordingly. In this paper, we propose to encode the look-up table of any function in a polynomial vector, whose coefficients can hold more data. The corresponding representation of the additive group Zq used in the RGSWbased bootstrapping is the group of monic monomial permutation matrices, which integrates the permutation matrix representation used by Alperin-Sheriffand Peikert in 2014, and the monic monomial representation used in the FHEW/TFHE scheme. We make comprehensive investigation of the new representation, and propose a new bootstrapping algorithm based on it. The new algorithm supports functional bootstrapping of large-plaintexts, and achieves polynomial reduction in key sizes and a constant-factor improvement in run-time cost.
Source: pdf
Extracted details¶
| Field | Value | Source | Confidence |
|---|---|---|---|
| contribution | Look-up table encoding in a polynomial vector for large-plaintext functional bootstrapping | llm | 0.80 |
| contribution | Monic monomial permutation matrix representation of Z_q integrating Alperin-Sheriff-Peikert and FHEW/TFHE representations | llm | 0.80 |
| contribution | BootMMPM bootstrapping algorithm with polynomial reduction in bootstrapping and key-switching key sizes | llm | 0.80 |
| contribution | Constant-factor run-time improvement over TFHE for large plaintexts, demonstrated in PALISADE | llm | 0.80 |
| schemes | TFHE | llm | 0.80 |
| schemes | FHEW | llm | 0.80 |
| schemes | RGSW | llm | 0.80 |
| libraries | PALISADE | llm | 0.80 |
| tasks | functional bootstrapping | llm | 0.80 |
| tasks | homomorphic look-up table evaluation | llm | 0.80 |
| hardware | Intel Core i7-10700 CPU (16 GB RAM) | llm | 0.80 |
| hardware | Intel Xeon Gold 6258R CPU (1 TB RAM) | llm | 0.80 |
| security level | Platform-2 experiments below 128-bit (both schemes compared at the same level); BootMMPM security lower than TFHE when r > 1 | llm | 0.80 |
| interaction | non-interactive FHE | llm | 0.80 |
| limitations | Security level of BootMMPM under the tested parameters is lower than TFHE's when r > 1 | llm | 0.80 |
| limitations | Key-packing optimizations that would shrink keys further were not implemented in the experiments | llm | 0.80 |
| limitations | Non-nega-cyclic functions require two bootstrapping rounds | llm | 0.80 |
| llm relevance | low | llm | 0.80 |
| key result | bootstrapping time: 2.64 s vs 10.09 s (TFHE) (7-bit plaintext, two bootstrapping rounds, Intel i7-10700, PALISADE v1.11.9) | llm | 0.80 |
| key result | speed-up over TFHE: ~3x (7-bit plaintext, PC platform) | llm | 0.80 |
| key result | bootstrapping time: 96.1 s vs 123.5 s (TFHE) (11-bit plaintext, Xeon Gold 6258R, log n = 10) | llm | 0.80 |
| key result | key size: constant 512 MB bootstrapping keys + 4.69 GB key-switching keys (BootMMPM, plaintext 5-11 bits; TFHE grows to 32 GB + 300 GB) | llm | 0.80 |
Keyword mentions¶
Counts of vocabulary terms in the full text (first page of occurrence in brackets).
- scheme: TFHE 41 [p.1], FHEW 26 [p.1], BFV 3 [p.2], GSW 2 [p.2]
- library: PALISADE 3 [p.4]
- model: Decision tree 1 [p.2]
- hardware: Intel Xeon 1 [p.20]
- technique: Bootstrapping 95 [p.1], Functional bootstrapping 33 [p.1], Key switching 16 [p.4], Lookup table 9 [p.1], Rotations 2 [p.4], Compiler 2 [p.19]
- assumption: RLWE 26 [p.2], LWE 7 [p.2], Lattice-based 2 [p.23]
- topic: Homomorphic encryption 18 [p.1]
Related papers¶
- this paper follow up of Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption — follow-up listed for scheme "FHEW with small keys (LMKCDEY)"
Files¶
| Path (under papers/) | Collection | Pages | Size | SHA-256 | Note |
|---|---|---|---|---|---|
references/schemes/s23-2_eprint-2025-1717.pdf |
schemes | 28 | 510.2 KB | 26b1e326351f |
Database key: large-plaintext-functional-bootstrapping-in-fhe-with-small · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set large-plaintext-functional-bootstrapping-in-fhe-with-small COLUMN VALUE