Search papers, labs, and topics across Lattice.
This paper introduces a logic-locking scheme utilizing the binary ring learning with error (bin-RLWE) algorithm, specifically designed for RISC-V system-on-chip (SoC) architectures. By securing critical logic paths during privilege switching without disclosing sensitive locking parameters, the method enhances security while maintaining a manageable overhead of 6.0% in LUTs and 6.9% in registers. The implementation achieves an unlock time of approximately 2.6 碌s, striking a balance between security and performance in system-level operations.
Securing critical logic paths without exposing sensitive parameters could redefine the landscape of logic-locking in SoC designs.
This study presents a logic-locking scheme based on the binary ring learning with error (bin-RLWE) algorithm, implemented in a reduced instruction set computer-five (RISC-V) system-on-chip (SoC) design. Unlike traditional logic-locking methods that require providing users with raw locking parameters, the proposed approach secures critical logic paths in the privilege switching process without exposing these sensitive parameters. The implemented locking module itself consumes 3519 lookup tables (LUTs) and 2645 registers, leading to an overall overhead of 6.0% in LUTs and 6.9% in registers compared to the baseline system. The unlock process requires about $2.6~\mu $ s, introducing moderate performance impact and primarily affecting system-level operations while preserving user-level computational efficiency.
