Shoumeng Yan
Research Subjects
Shoumeng Yan has a total of 160 co-authors. Here are the top co-authors by number of shared publications:
- Hongliang Tian
Ant Group, Hangzhou, China
7 shared publications - Fengwei Zhang
Southern University of Science and Technology, Shenzhen, China
6 shared publications - Guang Fan
Ant Group, China
6 shared publications - Mingzhe Zhang
Institute of Information Engineering, CAS
6 shared publications - Yilan Zhu
Ant Group, China
5 shared publications - Chenxu Wang
Research Institute of Trustworthy Autonomous Systems, Southern University of Science and Technology, Shenzhen, Guangdong, China; Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen, Guangdong, China and Department of Computing, The Hong Kong Polytechnic University, Kowloon, Hong Kong, China
4 shared publications
Latest Publications
- Open Access
- 5 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
PLUGSCHED+: Efficient and General Live Update and Recovery for Linux Scheduler
The scheduler is a critical component of the operating system (OS) and is tightly coupled with Linux. Production-level clouds often host various workloads, and these workloads require different schedulers to achieve high performance. Furthermore, the scheduler may encounter faults during runtime, leading to the kernel panic. Thus the capability of updating and recovering the scheduler lively without rebooting the OS is crucial for the production environments. However, existing techniques only apply for the fine-grained function-level updates, or require extra constraints such as virtualization when performing update and recovery. We therefore propose Plugsched+ to enable scheduler live update and recovery with two key novelties. First, with the idea of modularization, Plugsched+ decouples the scheduler from the Linux kernel to be an independent module. Second, Plugsched+ uses the data rebuild technique to migrate the state from the old scheduler to the new one during live update, or clear anomalous data during live recovery. This scheme can be directly applied to the Linux kernel scheduler in production environments without replacing the kernel. Experimental and production results show that Plugsched+ can effectively update and recover the scheduler lively, and the downtime is no more than tens of milliseconds.
- 5 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
Open AccessPLUGSCHED+: Efficient and General Live Update and Recovery for Linux Scheduler
The scheduler is a critical component of the operating system (OS) and is tightly coupled with Linux. Production-level clouds often host various workloads, and these workloads require different schedulers to achieve high performance. Furthermore, the scheduler may encounter faults during runtime, leading to the kernel panic. Thus the capability of updating and recovering the scheduler lively without rebooting the OS is crucial for the production environments. However, existing techniques only apply for the fine-grained function-level updates, or require extra constraints such as virtualization when performing update and recovery. We therefore propose Plugsched+ to enable scheduler live update and recovery with two key novelties. First, with the idea of modularization, Plugsched+ decouples the scheduler from the Linux kernel to be an independent module. Second, Plugsched+ uses the data rebuild technique to migrate the state from the old scheduler to the new one during live update, or clear anomalous data during live recovery. This scheme can be directly applied to the Linux kernel scheduler in production environments without replacing the kernel. Experimental and production results show that Plugsched+ can effectively update and recover the scheduler lively, and the downtime is no more than tens of milliseconds.
- Open Access
- 1 October 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Software Engineering
- Vol. 3 (ISSTA) , 3294-3315
When Optimizations Backfire: The Paradox of Plaintext Optimizations in Privacy-Preserving ML Compilers
Growing concerns about data security and privacy have fueled the widespread adoption of Privacy-Preserving Machine Learning (PPML). Cryptography-based PPML, which allows computation directly on encrypted data, significantly mitigates data leakage risks. To facilitate its adoption, Ciphertext Machine Learning Compilers (CMLCs) automate the translation of high-level ML procedures into low-level circuits for encrypted data. Recently, CMLCs have increasingly adopted infrastructure from Plaintext Machine Learning Compilers (PMLCs). While integrating Plaintext-Domain Optimization Passes (PDOPs) into CMLCs offers potential performance, usability, and extensibility benefits, our study shows it can also lead to severe performance regressions — a risk that has been largely overlooked. To address this, we introduce Hopta, a hybrid domain optimization defects tester and analyzer, aiming to find optimization bugs that can degrade the performance of compiled circuits from CMLCs. We carefully design two core components: (1) an optimization pipeline mutation mechanism to detect optimization anomalies by selectively enabling/disabling PDOPs, and (2) a profile-guided code reduction tool that efficiently simplifies defect-triggering programs to isolate optimization anomalies and facilitate debugging. Applying Hopta to SecretFlow-SPU, a production-grade CMLC, we identified 11 optimization defects (spanning matrix indexing,arithmetic/boolean conversion, and cost model deviation) that led to substantial performance regressions, with compiled circuits incurring up to 310.2% increased cost. Our comprehensive analysis provides empirical insights into fundamental differences between plaintext and ciphertext domain optimization strategies, offering crucial guidance for future CMLC development. We conclude with a brief discussion of extensions to other hybrid ML compilers, underscoring the methodology’s compiler-agnostic nature. This work establishes a new research direction for enhancing CMLC performance and the practical deployment of privacy-preserving ML systems.
- 1 October 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Software Engineering
- Vol. 3 (ISSTA) , 3294-3315
Open AccessWhen Optimizations Backfire: The Paradox of Plaintext Optimizations in Privacy-Preserving ML Compilers
Growing concerns about data security and privacy have fueled the widespread adoption of Privacy-Preserving Machine Learning (PPML). Cryptography-based PPML, which allows computation directly on encrypted data, significantly mitigates data leakage risks. To facilitate its adoption, Ciphertext Machine Learning Compilers (CMLCs) automate the translation of high-level ML procedures into low-level circuits for encrypted data. Recently, CMLCs have increasingly adopted infrastructure from Plaintext Machine Learning Compilers (PMLCs). While integrating Plaintext-Domain Optimization Passes (PDOPs) into CMLCs offers potential performance, usability, and extensibility benefits, our study shows it can also lead to severe performance regressions — a risk that has been largely overlooked. To address this, we introduce Hopta, a hybrid domain optimization defects tester and analyzer, aiming to find optimization bugs that can degrade the performance of compiled circuits from CMLCs. We carefully design two core components: (1) an optimization pipeline mutation mechanism to detect optimization anomalies by selectively enabling/disabling PDOPs, and (2) a profile-guided code reduction tool that efficiently simplifies defect-triggering programs to isolate optimization anomalies and facilitate debugging. Applying Hopta to SecretFlow-SPU, a production-grade CMLC, we identified 11 optimization defects (spanning matrix indexing,arithmetic/boolean conversion, and cost model deviation) that led to substantial performance regressions, with compiled circuits incurring up to 310.2% increased cost. Our comprehensive analysis provides empirical insights into fundamental differences between plaintext and ciphertext domain optimization strategies, offering crucial guidance for future CMLC development. We conclude with a brief discussion of extensions to other hybrid ML compilers, underscoring the methodology’s compiler-agnostic nature. This work establishes a new research direction for enhancing CMLC performance and the practical deployment of privacy-preserving ML systems.
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Towards Practical Live Migration for Heterogeneous Confidential Virtual Machines
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessTowards Practical Live Migration for Heterogeneous Confidential Virtual Machines
- 2 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. PP (02780070) , 1
OmniCifro: A Heterogeneous Acceleration System with PIM for Encrypted Computation
Fully Homomorphic Encryption (FHE) offers unparalleled data security guarantees for privacy-sensitive applications, but its widespread adoption is hindered by prohibitively high computation and memory capacity overheads compared to conventional computation on plaintext data. These inefficiencies create critical performance bottlenecks, limiting FHE’s practicality in real-world scenarios. To enhance performance, we systematically characterize FHE workloads, noting that existing accelerators struggle with state-of-the-art FHE schemes due to bandwidth constraints and the diverse computation patterns of different FHE primitives. In this work, we propose OmniCifro, a heterogeneous FHE acceleration framework that addresses performance bottlenecks through architectural design and dataflow optimization. OmniCifro combines a specialized Homomorphic Processing Unit (HPU) for compute-intensive kernels with Processing-in-Memory (PIM) modules for memory-bound operations, featuring dedicated functional units for polynomial arithmetic and element-wise operations. We further optimize the kernel mapping strategy for advanced FHE algorithms to reduce communication overhead between HPU and PIM. Experimental evaluations demonstrate that OmniCifro achieves a 1.6× performance improvement over ARK.
- 2 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. PP (02780070) , 1
OmniCifro: A Heterogeneous Acceleration System with PIM for Encrypted Computation
Fully Homomorphic Encryption (FHE) offers unparalleled data security guarantees for privacy-sensitive applications, but its widespread adoption is hindered by prohibitively high computation and memory capacity overheads compared to conventional computation on plaintext data. These inefficiencies create critical performance bottlenecks, limiting FHE’s practicality in real-world scenarios. To enhance performance, we systematically characterize FHE workloads, noting that existing accelerators struggle with state-of-the-art FHE schemes due to bandwidth constraints and the diverse computation patterns of different FHE primitives. In this work, we propose OmniCifro, a heterogeneous FHE acceleration framework that addresses performance bottlenecks through architectural design and dataflow optimization. OmniCifro combines a specialized Homomorphic Processing Unit (HPU) for compute-intensive kernels with Processing-in-Memory (PIM) modules for memory-bound operations, featuring dedicated functional units for polynomial arithmetic and element-wise operations. We further optimize the kernel mapping strategy for advanced FHE algorithms to reduce communication overhead between HPU and PIM. Experimental evaluations demonstrate that OmniCifro achieves a 1.6× performance improvement over ARK.
- 4 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (8) , 2857-2869
Keystone-Vault: Hardware-Assisted Efficient Intra-Enclave Isolation for RISC-V TEEs
RISC-V’s open architecture provides an ideal platform for customizable security features, particularly in trusted execution environments (TEEs). However, existing RISC-V TEEs such as Keystone adopt a monolithic model where applications run with all dependencies in a single isolation domain, leading to bloated trusted computing bases (TCBs). When trusted applications grow large, vulnerabilities in any third-party library can compromise the entire enclave’s security. To address these challenges, we propose Keystone-Vault, which enables fine-grained isolation within RISC-V enclaves. Keystone-Vault reimplements and improves the Donky memory protection key mechanism and introduces a hardware-software co-designed isolation framework. The core idea is to partition each enclave into multiple regions, called Vaults, to isolate specific portions of code and data with hardware-assisted access control. Mean-while, our lightweight vault-switching mechanism runs entirely in user mode, making it more lightweight than existing RISC-V TEE designs that rely on Physical Memory Protection (PMP) to enforce security-domain partitioning. Proof-of-concept (PoC) experiments based on a prototype of Keystone-Vault implemented on the QEMU emulation platform demonstrate that Keystone-Vault successfully defends against attacks from a malicious OS and confines security risks from untrusted libraries within their respective vaults. Evaluation on realistic and complex workloads further demonstrates that, compared with baseline Keystone, Keystone-Vault introduces only 3%–7% additional performance overhead.
- 4 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (8) , 2857-2869
Keystone-Vault: Hardware-Assisted Efficient Intra-Enclave Isolation for RISC-V TEEs
RISC-V’s open architecture provides an ideal platform for customizable security features, particularly in trusted execution environments (TEEs). However, existing RISC-V TEEs such as Keystone adopt a monolithic model where applications run with all dependencies in a single isolation domain, leading to bloated trusted computing bases (TCBs). When trusted applications grow large, vulnerabilities in any third-party library can compromise the entire enclave’s security. To address these challenges, we propose Keystone-Vault, which enables fine-grained isolation within RISC-V enclaves. Keystone-Vault reimplements and improves the Donky memory protection key mechanism and introduces a hardware-software co-designed isolation framework. The core idea is to partition each enclave into multiple regions, called Vaults, to isolate specific portions of code and data with hardware-assisted access control. Mean-while, our lightweight vault-switching mechanism runs entirely in user mode, making it more lightweight than existing RISC-V TEE designs that rely on Physical Memory Protection (PMP) to enforce security-domain partitioning. Proof-of-concept (PoC) experiments based on a prototype of Keystone-Vault implemented on the QEMU emulation platform demonstrate that Keystone-Vault successfully defends against attacks from a malicious OS and confines security risks from untrusted libraries within their respective vaults. Evaluation on realistic and complex workloads further demonstrates that, compared with baseline Keystone, Keystone-Vault introduces only 3%–7% additional performance overhead.
- Open Access
- 26 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 311-327
Pyramid: A Secure, Resource-Efficient, and Pluggable Kubernetes for Multi-Tenancy
- 26 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 311-327
Open AccessPyramid: A Secure, Resource-Efficient, and Pluggable Kubernetes for Multi-Tenancy
- Open Access
- 22 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 25-40
A Framework for Developing and Optimizing Fully Homomorphic Encryption Programs on GPUs
- 22 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 25-40
Open AccessA Framework for Developing and Optimizing Fully Homomorphic Encryption Programs on GPUs
- Open Access
- 22 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 715-731
Falcon: Algorithm-Hardware Co-Design for Efficient Fully Homomorphic Encryption Accelerator
Fully homomorphic encryption (FHE) enables computation on encrypted data without compromising privacy, positioning it as a promising solution for secure cloud computing. However, its substantial computational overhead impedes practical deployment, prompting the development of dedicated hardware accelerators. In practice, when deploying cryptographic algorithm optimizations on FHE accelerators, hardware constraints typically such as limited memory capacity, often lead to a disparity between theoretical algorithmic advantage and achievable hardware efficiency. We present Falcon, an algorithm-hardware co-design for efficient FHE acceleration. We first conduct computational analysis of the prevalent minimum key-switching method, and propose hardware-oriented algorithmic optimizations that substantially reduce computational overhead. We then introduce further refinements to decrease the inherent inter-cluster communication overhead in multi-cluster architectures, while preserving computational advantages. By characterizing the arithmetic patterns intrinsic to primary primitives, we devise hardware-specific arithmetic fusion and functional reuse across computational units. Finally, we present a memory-adaptive strategy to effectively deploy Falcon to prior accelerators under constrained on-chip memory budgets. Experimental results demonstrate that applying our algorithm–hardware co-design delivers up to a 1.48× speedup with a negligible area overhead (only a 0.8% increase).
- 22 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 715-731
Open AccessFalcon: Algorithm-Hardware Co-Design for Efficient Fully Homomorphic Encryption Accelerator
Fully homomorphic encryption (FHE) enables computation on encrypted data without compromising privacy, positioning it as a promising solution for secure cloud computing. However, its substantial computational overhead impedes practical deployment, prompting the development of dedicated hardware accelerators. In practice, when deploying cryptographic algorithm optimizations on FHE accelerators, hardware constraints typically such as limited memory capacity, often lead to a disparity between theoretical algorithmic advantage and achievable hardware efficiency. We present Falcon, an algorithm-hardware co-design for efficient FHE acceleration. We first conduct computational analysis of the prevalent minimum key-switching method, and propose hardware-oriented algorithmic optimizations that substantially reduce computational overhead. We then introduce further refinements to decrease the inherent inter-cluster communication overhead in multi-cluster architectures, while preserving computational advantages. By characterizing the arithmetic patterns intrinsic to primary primitives, we devise hardware-specific arithmetic fusion and functional reuse across computational units. Finally, we present a memory-adaptive strategy to effectively deploy Falcon to prior accelerators under constrained on-chip memory budgets. Experimental results demonstrate that applying our algorithm–hardware co-design delivers up to a 1.48× speedup with a negligible area overhead (only a 0.8% increase).