Skip to content

XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering

Year: 2025

type: technique collection: schemes status: auto

Links: ePrint 2025/74

Summary

Auto-extracted summary (unreviewed)

XBOOT builds on the BLEACH approach for exact Boolean computation in CKKS and makes XOR gates almost free by evaluating them as plain additions (lazy reduction) and absorbing the resulting overflows during CKKS bootstrapping through careful handling of the modulus chain and scale factors, so an XOR consumes no multiplicative depth. Implemented in Lattigo v6 at 128-bit security with ring degree 2^15, AES-CKKS transciphering reaches 256 KB per 236 s (14.4 ms amortized per block), a 5x latency and 2.5x throughput improvement over Aharoni et al., and Rasta-CKKS transciphering exceeds 2 KB/s, more than 10x faster than without XBOOT.

Abstract

The CKKS scheme is traditionally recognized for approximate homomorphic encryption of real numbers, but BLEACH (Drucker et al., JoC 2024) extends its capabilities to handle exact computations on binary or small integer numbers. Despite this advancement, BLEACH’s approach of simulating XOR gates via (a −b)2 incurs one multiplication per gate, which is computationally expensive in homomorphic encryption. To this end, we introduce XBOOT, a new framework built upon BLEACH’s blueprint but allows for almost free evaluation of XOR gates. The core concept of XBOOT involves lazy reduction, where XOR operations are simulated with the less costly addition operation, a+b, leaving the management of potential overflows to later stages. We carefully handle the modulus chain and scale factors to ensure that the overflows are managed during the CKKS bootstrapping phase, preserving the correct XOR result without extra cost. We use AES-CKKS transciphering as a benchmark to test the capability of XBOOT, and achieve a throughput exceeding one kilobyte per second, which represents a 2.5× improvement over the state-of-the-art (Aharoni et al., HES 2023). Moreover, XBOOT enables the practical execution of tasks with extensive XOR operations that were previously challenging for CKKS. For example, we can do Rasta-CKKS transciphering at over two kilobytes per second, more than 10× faster than the baseline without XBOOT.
Source: pdf

Extracted details

Field Value Source Confidence
contribution XBOOT framework for near-free XOR gates in CKKS via lazy reduction (XOR simulated as addition) llm 0.90
contribution modulus-chain and scale-factor management so that overflow cleanup happens inside CKKS bootstrapping at no extra cost llm 0.90
contribution reduced parameters for AES-CKKS transciphering (N = 2^15, log(QP) <= 830 bits) with sparse-secret encapsulation llm 0.90
contribution application of XBOOT to Rasta-CKKS transciphering llm 0.90
contribution open-source implementation in Lattigo v6 and comparison with BGV- and TFHE-based AES transciphering llm 0.90
schemes CKKS llm 0.90
schemes AES llm 0.90
schemes Rasta llm 0.90
libraries Lattigo llm 0.90
tasks transciphering / hybrid homomorphic encryption llm 0.90
tasks Boolean circuit evaluation in CKKS llm 0.90
hardware Intel Xeon Platinum 8369B 64-core 2.50 GHz CPU, 128 GB RAM llm 0.90
security level 128-bit llm 0.90
threat model client-server (client encrypts with AES/Rasta, server transciphers into CKKS) llm 0.90
interaction non-interactive FHE llm 0.90
code url https://github.com/AntCPLab/xboot-ckks-transciphering.1.2 (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/AntCPLab/xboot-ckks-transciphering llm 0.90
limitations TFHE-based AES transciphering methods achieve lower latency because they process one block at a time; CKKS methods win only on amortized throughput llm 0.90
limitations the large FHE memory footprint limits parallelization: 64 threads give only a 20-30x speedup llm 0.90
llm relevance low llm 0.90
key result AES transciphering latency: 236 s for 2^14 blocks (256 KB) (Param-AES-14, N = 2^15, 64 threads) llm 0.90
key result AES transciphering amortized time per block: 14.4 ms (vs 37 ms for Aharoni et al.) (Param-AES-14, 64 threads) llm 0.90
key result AES transciphering with smaller parameters: 153 s latency, 18.7 ms per block, 2^13 blocks (Param-AES-13, N = 2^14, 64 threads) llm 0.90
key result improvement over state-of-the-art CKKS AES transciphering: 5x latency, 2.5x throughput (vs Aharoni et al. (HES 2023)) llm 0.90
key result Rasta-CKKS transciphering throughput: > 2 KB/s, more than 10x over baseline without XBOOT (Param-Rasta-14) llm 0.90
key result multi-thread scaling: 20-30x with 64 threads over single thread (authors' implementation) llm 0.90

Keyword mentions

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

  • scheme: CKKS 88 [p.1], Hybrid HE / transciphering 72 [p.1], TFHE 21 [p.2], BGV 8 [p.2], MPC 5 [p.2], BFV 4 [p.2], RSA 4 [p.23], FHEW 2 [p.24]
  • library: Lattigo 3 [p.17], Concrete (Zama) 1 [p.18]
  • hardware: Intel Xeon 1 [p.17]
  • technique: Bootstrapping 58 [p.1], Lookup table 18 [p.6], SIMD packing 8 [p.3], Rescaling 4 [p.5], Key switching 3 [p.5], Relinearization 3 [p.5], Polynomial approximation 3 [p.5], Matrix multiplication 3 [p.5], Convolution 3 [p.25], Rotations 2 [p.5], RNS 2 [p.4]
  • assumption: LWE 2 [p.21]
  • task: Genomics 3 [p.21], Language modeling / generation 2 [p.9]
  • topic: Homomorphic encryption 51 [p.1], Open source 4 [p.3], Benchmark 2 [p.1]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/schemes/s10-1_eprint-2025-074.pdf schemes 27 781.6 KB 6275810a57f1

Database key: xboot-free-xor-gates-for-ckks-with-applications-to-transciphering · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set xboot-free-xor-gates-for-ckks-with-applications-to-transciphering COLUMN VALUE