Skip to content

Bootstrapping GBFV with CKKS

Year: 2025

type: technique collection: schemes status: auto

Links: ePrint 2025/888

Summary

Auto-extracted summary (unreviewed)

Proposes a bootstrapping method for Generalized BFV (GBFV, which covers BFV and CLPX with polynomial plaintext moduli) that treats the GBFV ciphertext noise as a CKKS plaintext and refreshes it with CKKS bootstrapping, using a new noise extraction and recovery procedure for polynomial plaintext moduli. Built on Lattigo, it bootstraps large-precision GBFV and CLPX instances for the first time, handling plaintext moduli above 500,000 bits in under 20 s, and adds GBFV-to-CKKS and CKKS-to-GBFV conversions for non-arithmetic operations.

Abstract

Generalized BFV (GBFV) replaces the integer plaintext modulus of BFV with a polynomial plaintext modulus, yielding a flexible tradeoff between precision and SIMD parallelism. This framework encompasses both BFV and CLPX, with CLPX corresponding to the extreme large-precision endpoint where very large plaintext moduli can be supported without increasing the underlying RLWE dimension. However, bootstrapping remains a bottleneck throughout the large-precision GBFV regime: existing GBFV bootstrapping methods rely on BFV-style linear transformations, whose noise growth becomes prohibitive as the plaintext precision grows. We introduce a GBFV bootstrapping method that treats the noise of a GBFV ciphertext as a CKKS plaintext and bootstraps it through CKKS. The main obstacle is that the BFV-from- CKKS bootstrapping technique of Kim et al. (CCS’24) relies on integer plaintext scaling and does not directly apply to GBFV, whose scaling factor involves a polynomial plaintext modulus. We overcome this by developing a noise-extraction and recovery procedure tailored to polynomial plaintext moduli. Our implementation bootstraps large-precision GBFV instances, including CLPX for the first time, supporting plaintext moduli of more than 500,000 bits in less than 20 seconds. We also introduce GBFV-to-CKKS and CKKS-to-GBFV conversions, interpreting the former as an approximate digit-decomposition toolkit for non-arithmetic operations over large GBFV plaintext spaces.
Source: pdf

Extracted details

Field Value Source Confidence
contribution GBFV bootstrapping via CKKS bootstrapping of the ciphertext noise llm 0.90
contribution noise-extraction and recovery procedure tailored to polynomial plaintext moduli llm 0.90
contribution first bootstrapping for CLPX (large-precision endpoint of GBFV) llm 0.90
contribution GBFV-to-CKKS and CKKS-to-GBFV conversions as an approximate digit-decomposition toolkit llm 0.90
contribution proof-of-concept implementation in Lattigo with public code llm 0.90
schemes GBFV llm 0.90
schemes BFV llm 0.90
schemes CKKS llm 0.90
schemes CLPX llm 0.90
libraries Lattigo llm 0.90
tasks FHE bootstrapping llm 0.90
tasks large-precision integer arithmetic on encrypted data llm 0.90
hardware Apple M4 Max, 128 GB RAM, single-threaded llm 0.90
security level 128-bit llm 0.90
interaction non-interactive FHE llm 0.90
code url https://github.com/jaehyungkim0/GBFV-Boot-CKKS-1 (URL found in the full text) regex 0.50
code url https://github.com/jaehyungkim0/GBFV-Boot-CKKS-2 (URL found in the full text) regex 0.50
code url https://github.com/tuneinsight/lattigo (URL found in the full text) regex 0.50
code url https://github.com/jaehyungkim0/GBFV-Boot-CKKS-1 llm 0.90
limitations for small plaintext moduli such as p = 2^16 + 1 the existing BFV-style GBFV bootstrapping remains preferable llm 0.90
limitations denoising factor is much smaller than the N = 2^15 implementation of prior work unless more META-BTS iterations are used llm 0.90
limitations proof-of-concept implementation llm 0.90
llm relevance low llm 0.90
key result GBFV bootstrapping latency: about 18.4 s (N = 2^16, log QP = 1288, across all tested (b, k) parameter choices) llm 0.90
key result supported plaintext modulus size: 2^19 = 524,288 bits (CLPX setting k = 1, b = 256, bootstrapped in 18.4 s) llm 0.90
key result denoising factor: about 70.8-77.4 bits (per bootstrap, CKKS bootstrapping with ~80 bits precision via META-BTS) llm 0.90
key result latency vs prior GBFV bootstrapping [30]: prior work about 7.4x lower latency and 3.7x lower amortized latency (small plaintext modulus p = 2^16 + 1, normalized denoising factor) llm 0.90

Keyword mentions

Counts of vocabulary terms in the full text (first page of occurrence in brackets).

  • scheme: CKKS 201 [p.1], BFV 51 [p.1], BGV 5 [p.1], TFHE 3 [p.8], Zero-knowledge proofs 2 [p.2]
  • library: HElib 1 [p.33], Lattigo 1 [p.30], MP-SPDZ 1 [p.2]
  • model: GPT 1 [p.33]
  • hardware: Apple silicon 1 [p.25]
  • technique: Bootstrapping 150 [p.1], Rescaling 12 [p.4], NTT 8 [p.15], SIMD packing 6 [p.1], RNS 6 [p.15], Modulus switching 5 [p.15]
  • assumption: RLWE 14 [p.1]
  • topic: Homomorphic encryption 27 [p.1], Benchmark 3 [p.8], Open source 3 [p.25]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/schemes/s17-1_eprint-2025-888.pdf schemes 37 582.9 KB 76dd61d4821d

Database key: bootstrapping-gbfv-with-ckks · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set bootstrapping-gbfv-with-ckks COLUMN VALUE