This book provides an introduction to the key concepts of Fully Homomorphic Encryption (FHE)-based computing, and discusses the challenges associated with architecting FHE-based computing systems. Readers will see that due to FHE’s ability to compute on encrypted data, it is a promising solution to address privacy concerns arising from cloud-based services commonly used for a variety of applications including healthcare, financial, transportation, and weather forecasting. This book explains the fundamentals of the FHE operations and then presents an architectural analysis of the FHE-based computing. The authors also highlight challenges associated with accelerating FHE on various commodity platforms and argue that the FPGA platform provides a sweet spot in making privacy-preserving computing plausible.
Author(s): Rashmi Agrawal, Ajay Joshi
Series: Synthesis Lectures on Computer Architecture
Edition: 1
Publisher: Springer
Year: 2023
Language: English
Commentary: Publisher PDF
Pages: 87
City: Cham
Tags: Privacy-Preserving Computing; Homomorphic Encryption; Machine Learning On Private Data; CKKS Scheme; Compute Challenges In FHE
Preface
Acknowledgements
Contents
1 Introduction
1.1 Privacy-Preserving Computing
1.2 Homomorphic Encryption
1.3 A Brief History of HE
1.3.1 First-Generation HE Schemes
1.3.2 Second-Generation HE Schemes
1.3.3 Third-Generation HE Schemes
1.4 Alternate Cryptographic Technique for PPC
1.4.1 Multi-party Computation
1.4.2 HE Versus MPC
1.5 Non-cryptographic Techniques for PPC
1.5.1 Anonymization
1.5.2 Differential Privacy
1.5.3 HE Versus Non-cryptographic PPC Techniques
1.6 Summary
2 The CKKS FHE Scheme
2.1 CKKS Parameters
2.2 Basic Compute Optimizations in CKKS Scheme
2.2.1 Residue Number System (RNS)
2.2.2 Modular Arithmetic
2.2.3 NTT/iNTT
2.3 Key Generation
2.3.1 Public Keys
2.3.2 Switching Keys
2.4 Client-Side Operations
2.4.1 Encoding
2.4.2 Encryption
2.4.3 Decryption
2.4.4 Decoding
2.4.5 Example of Encode and Decode Operation
2.5 Server-Side Operations
2.5.1 Addition
2.5.2 Multiplication
2.5.3 Rescale
2.5.4 Rotate
2.5.5 Conjugate
2.5.6 Key Switching
2.5.7 Noise Growth
2.5.8 Bootstrapping
2.6 Example of an FHE-Based Computing with CKKS Scheme
2.7 Summary
3 Architectural Analysis of CKKS FHE Scheme
3.1 Compute Bottlenecks
3.1.1 Modular Arithmetic Operations
3.1.2 NTT/iNTT Operations
3.1.3 Choice of Parameters
3.2 Memory Bottlenecks
3.2.1 On-chip Memory Size
3.2.2 Main Memory
3.2.3 Changing Data Access Pattern
3.2.4 Parameters
3.3 Arithmetic Intensity Analysis of Basic Operations
3.4 Arithmetic Intensity Analysis of Bootstrapping
3.5 Arithmetic Intensity Analysis of LR Model Training
3.6 Summary
4 Designing Computing Systems for CKKS FHE Scheme
4.1 CPU-Based Designs
4.1.1 Compute Versus Memory Trade-off
4.2 GPU-Based Designs
4.2.1 Compute Versus Memory Trade-off
4.3 FPGA-Based Designs
4.3.1 Compute Versus Memory Trade-off
4.4 ASIC-Based Designs
4.4.1 Compute Versus Memory Trade-off
4.5 Summary
5 Summary
5.1 Future Perspectives
5.1.1 Increase Main Memory Bandwidth with Improved Utilization
5.1.2 Use In-Memory/Near-Memory Computing/Wafer-Scale Systems
5.1.3 Future Improvements to FHE Schemes
