Makoto Ikeda
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Makoto Ikeda has a total of 74 co-authors. Here are the top co-authors by number of shared publications:
- Kunihiro Asada
VLSI Design and Education Center (VDEC), The University of Tokyo
10 shared publications - Kunihiro Asada
VLSI Design and Education Center, The University of Tokyo, Tokyo 113-0032, Japan
3 shared publications - Anawin Opasatian
Department of Electrical Engineering and Information Systems, Graduate School of Engineering, The University of Tokyo, Tokyo, Japan
2 shared publications Hidehiro Takata
Design Platform Development Division, Renesas Electronics Corporation, University of Tokyo 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
2 shared publicationsJinmyoung Kim
Dept. of Electronic Engineering, University of Tokyo
2 shared publications- Kunihiro Asada
Dept. of Electronic Engineering, University of Tokyo Tokyo, Japan; VLSI Design and Education Center (VDEC), University of Tokyo, Tokyo, Japan
2 shared publications
Latest Publications
- Open Access
- 2 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-3
Designing Chips in the Browser: Democratizing ASIC Prototyping with Gate Arrays
- 2 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-3
Open AccessDesigning Chips in the Browser: Democratizing ASIC Prototyping with Gate Arrays
- Open Access
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
- Vol. E109.A (8) , 1359-1366
Error Rate Analysis and Hardware Implementation Assessment of Binary Ring Learning with Errors
This paper proposes an error rate analysis method for Binary Ring Learning with Errors (bin-RLWE) cryptosystems where the secret and error are binary. We derive a computable method for the bit error rate (BER) in encryption and decryption processes using a normal approximation, and estimate an upper bound on the overall decryption failure probability (at least one bit error) via a union bound. The framework was validated through 109 full encrypt-decrypt tests for each of four parameter sets. Based on this analysis, we provide parameter selections for implementations requiring overall failure probability bounds of 2-16 and 2-32. We also evaluate the hardware performance of these parameters on 28 nm ASIC and Xilinx Kintex-7 FPGA platforms, measuring resource usage and latency. Comparisons with conventional RLWE implementations show that bin-RLWE achieves better resource efficiency and faster processing. Unlike Number Theoretic Transform-based RLWE implementations that commonly constrain dimensions in practice, our shifting convolution-based approach removes this implementation constraint, enabling more flexible parameter choices. Finally, we discuss the correctness-security trade-off in bin-RLWE and report estimated security levels against primal attack for the selected parameter sets.
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
- Vol. E109.A (8) , 1359-1366
Open AccessError Rate Analysis and Hardware Implementation Assessment of Binary Ring Learning with Errors
This paper proposes an error rate analysis method for Binary Ring Learning with Errors (bin-RLWE) cryptosystems where the secret and error are binary. We derive a computable method for the bit error rate (BER) in encryption and decryption processes using a normal approximation, and estimate an upper bound on the overall decryption failure probability (at least one bit error) via a union bound. The framework was validated through 109 full encrypt-decrypt tests for each of four parameter sets. Based on this analysis, we provide parameter selections for implementations requiring overall failure probability bounds of 2-16 and 2-32. We also evaluate the hardware performance of these parameters on 28 nm ASIC and Xilinx Kintex-7 FPGA platforms, measuring resource usage and latency. Comparisons with conventional RLWE implementations show that bin-RLWE achieves better resource efficiency and faster processing. Unlike Number Theoretic Transform-based RLWE implementations that commonly constrain dimensions in practice, our shifting convolution-based approach removes this implementation constraint, enabling more flexible parameter choices. Finally, we discuss the correctness-security trade-off in bin-RLWE and report estimated security levels against primal attack for the selected parameter sets.
- Open Access
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34 (2) , 579-592
Design Space Exploration of a Unified FPGA Accelerator for Elliptic-Curve-Based Functions in Attribute-Based Encryption
Attribute-based encryption (ABE) requires several elliptic-curve-based functions, including elliptic-curve scalar multiplication (ECSM), hashing to the curve, and pairing. Although these computations share underlying similarities, prior hardware designs have shown that distinct architectures yield better performance for each function. Consequently, it remains unclear which architecture offers optimal performance when supporting all required functions within a unified design. In this work, we present a design space exploration methodology in which the design is parameterized using a defined set of design parameters, and an automatic schedule generator is employed to estimate the cycle counts for each function. This allows us to identify the configuration that minimizes latency for both individual functions and complete cryptographic operations. The versatility of our approach is demonstrated through two case studies: 1) ECSM over Curve25519, Secp256k1, NIST-P256, and NIST-P384; and 2) multiple elliptic-curve functions over the BLS12-381 curve. In the first case, the optimal configuration implemented on a Virtex7 field-programmable gate array (FPGA) achieves up to a 15% latency reduction compared to state-of-the-art designs. In the second case, the unified accelerator in its optimal configuration supports all core ABE functions and achieves superior latency or throughput-per-area (TPA), and in some cases both, compared to existing designs.
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34 (2) , 579-592
Open AccessDesign Space Exploration of a Unified FPGA Accelerator for Elliptic-Curve-Based Functions in Attribute-Based Encryption
Attribute-based encryption (ABE) requires several elliptic-curve-based functions, including elliptic-curve scalar multiplication (ECSM), hashing to the curve, and pairing. Although these computations share underlying similarities, prior hardware designs have shown that distinct architectures yield better performance for each function. Consequently, it remains unclear which architecture offers optimal performance when supporting all required functions within a unified design. In this work, we present a design space exploration methodology in which the design is parameterized using a defined set of design parameters, and an automatic schedule generator is employed to estimate the cycle counts for each function. This allows us to identify the configuration that minimizes latency for both individual functions and complete cryptographic operations. The versatility of our approach is demonstrated through two case studies: 1) ECSM over Curve25519, Secp256k1, NIST-P256, and NIST-P384; and 2) multiple elliptic-curve functions over the BLS12-381 curve. In the first case, the optimal configuration implemented on a Virtex7 field-programmable gate array (FPGA) achieves up to a 15% latency reduction compared to state-of-the-art designs. In the second case, the unified accelerator in its optimal configuration supports all core ABE functions and achieves superior latency or throughput-per-area (TPA), and in some cases both, compared to existing designs.
- 10 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 33 (7) , 2049-2053
Implementing Homomorphic Encryption-Based Logic Locking in System-On-Chip 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.
- 10 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 33 (7) , 2049-2053
Implementing Homomorphic Encryption-Based Logic Locking in System-On-Chip 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.
- Open Access
- 4 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 33 (1) , 114-127
Manipulated Lookup Table Method for Efficient High-Performance Modular Multiplier
Modular multiplication is a fundamental operation in many cryptographic systems, with its efficiency playing a crucial role in the overall performance of these systems. Since many cryptographic systems operate with a fixed modulus, we propose an enhancement to the fixed modulus lookup table (LuT) method used for modular reduction, which we refer to as the manipulated LuT (MLuT) method. Our approach applies to any modulus and has demonstrated comparable performance compared with some specialized reduction algorithms designed for specific moduli. The strength of our proposed method in terms of circuit performance is shown by implementing it on Virtex7 and Virtex Ultrascale+ FPGA as the LUT-based MLuT modular multiplier (LUT-MLuTMM) with generalized parallel counters (GPCs) used in the summation step. In one-stage implementations, our proposed method achieves up to a 90% reduction in area and a 50% reduction in latency compared with the generic LuT method. In multistage implementations, our approach offers the best area-interleaved time product, with improvements of 39%, 13%, and 29% over the current state-of-the-art for $\sim$ 256-bit, SIKE434, and BLS12-381 modular multipliers, respectively. These results demonstrate the potential of our method for high-performance cryptographic accelerators employing a fixed modulus.
- 4 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 33 (1) , 114-127
Open AccessManipulated Lookup Table Method for Efficient High-Performance Modular Multiplier
Modular multiplication is a fundamental operation in many cryptographic systems, with its efficiency playing a crucial role in the overall performance of these systems. Since many cryptographic systems operate with a fixed modulus, we propose an enhancement to the fixed modulus lookup table (LuT) method used for modular reduction, which we refer to as the manipulated LuT (MLuT) method. Our approach applies to any modulus and has demonstrated comparable performance compared with some specialized reduction algorithms designed for specific moduli. The strength of our proposed method in terms of circuit performance is shown by implementing it on Virtex7 and Virtex Ultrascale+ FPGA as the LUT-based MLuT modular multiplier (LUT-MLuTMM) with generalized parallel counters (GPCs) used in the summation step. In one-stage implementations, our proposed method achieves up to a 90% reduction in area and a 50% reduction in latency compared with the generic LuT method. In multistage implementations, our approach offers the best area-interleaved time product, with improvements of 39%, 13%, and 29% over the current state-of-the-art for $\sim$ 256-bit, SIKE434, and BLS12-381 modular multipliers, respectively. These results demonstrate the potential of our method for high-performance cryptographic accelerators employing a fixed modulus.
- Open Access
- 3 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 94008-94017
Toward Bootstrapping-Free Homomorphic Encryption-Based GRU Network for Text Classification
Homomorphic encryption (HE) is a promising method in privacy-preserving cloud computing. Applying HE on feedforwad neural networks has been frequently reported recently but the research on recurrent neural networks is still insufficient. In previous studies, HE-based GRU is built with bootstrapping due to the changeable input length and large number of required multiplications, which is not time-efficient. In this study, we give a guideline of building bootstrapping-free HE-based GRU for text classification tasks. We discuss the methods of pre-processing of texts to decrease the input sequence length but keep the accuracy in a comparable level as the original GRU. The architecture of GRU is designed with flexibility to process the input sequence with different lengths while fixing the recurrent steps. At last, the HE parameter selection is discussed.We analyze the noise raised from HE operations and select the parameter that ensure the results from encrypted data are the same as that on plaintexts. The proposed model is evaluated on 6 popular text datasets, and the results show that the accuracy is only lower than the original GRU by at most 4.2%. Despite the complicated calculations in GRU models, the model is light-weighted and the fastest inference in our implementations costs only 10 minutes.We show the potential of applying HE schemes on complex models without bootstrapping to achieve fast encrypted computations.
- 3 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 94008-94017
Open AccessToward Bootstrapping-Free Homomorphic Encryption-Based GRU Network for Text Classification
Homomorphic encryption (HE) is a promising method in privacy-preserving cloud computing. Applying HE on feedforwad neural networks has been frequently reported recently but the research on recurrent neural networks is still insufficient. In previous studies, HE-based GRU is built with bootstrapping due to the changeable input length and large number of required multiplications, which is not time-efficient. In this study, we give a guideline of building bootstrapping-free HE-based GRU for text classification tasks. We discuss the methods of pre-processing of texts to decrease the input sequence length but keep the accuracy in a comparable level as the original GRU. The architecture of GRU is designed with flexibility to process the input sequence with different lengths while fixing the recurrent steps. At last, the HE parameter selection is discussed.We analyze the noise raised from HE operations and select the parameter that ensure the results from encrypted data are the same as that on plaintexts. The proposed model is evaluated on 6 popular text datasets, and the results show that the accuracy is only lower than the original GRU by at most 4.2%. Despite the complicated calculations in GRU models, the model is light-weighted and the fastest inference in our implementations costs only 10 minutes.We show the potential of applying HE schemes on complex models without bootstrapping to achieve fast encrypted computations.
- Open Access
- 1 January 2024
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in IEICE Electronics Express
- Vol. 21 (12)
Advancing energy efficiency of spiking neural network accelerator via dynamic predictive early stopping
Spiking Neural Network (SNN) accelerators, recognized for their potential in neuromorphic processing, have yet to fully realize high energy efficiency, especially in widespread conventional non-event-based tasks, thus limiting their broader applicability. In response, we introduce an effective, cost-efficient method named dynamic predictive early stopping. This method enhances energy efficiency by predicting and stopping nearly 70% of non-essential computations during inference while ensuring near-lossless performance. The FPGA-based accelerator prototype demonstrated a 35% energy efficiency net improvement (to 208.74 GOPS/W, leading for its class). Simultaneously, the proposed method incorporates only minimal (<0.41%) extensions to hardware. With potential for further improvements and explorations, this method could bring substantial impact by being widely adopted by SNN accelerators.
- 1 January 2024
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in IEICE Electronics Express
- Vol. 21 (12)
Open AccessAdvancing energy efficiency of spiking neural network accelerator via dynamic predictive early stopping
Spiking Neural Network (SNN) accelerators, recognized for their potential in neuromorphic processing, have yet to fully realize high energy efficiency, especially in widespread conventional non-event-based tasks, thus limiting their broader applicability. In response, we introduce an effective, cost-efficient method named dynamic predictive early stopping. This method enhances energy efficiency by predicting and stopping nearly 70% of non-essential computations during inference while ensuring near-lossless performance. The FPGA-based accelerator prototype demonstrated a 35% energy efficiency net improvement (to 208.74 GOPS/W, leading for its class). Simultaneously, the proposed method incorporates only minimal (<0.41%) extensions to hardware. With potential for further improvements and explorations, this method could bring substantial impact by being widely adopted by SNN accelerators.
- 18 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Solid-State Circuits Magazine
- Vol. 15 (3) , 36-42
Reflections on Progress in Digital Circuits: From basics to untangle past trials
A digital circuit is a system that discretizes and processes information and mainly expresses numerical values in binary numbers of 1 and 0. Logical 1 s and 0 s are physically represented by high and low voltages, large and small currents (or no current), and so on. Integrated digital circuits, called RTL , were introduced by Fairchild in 1961. To prevent short currents among input parallel registers, parallel diodes were introduced, called DTL . TTL, which replaced input diodes with bipolar transistors, were patented by James Buie of Pacific Semiconductor in 1961 and became the most popular logic configuration of the next two decades. General Microelectronics introduced the first commercial PMOS digital circuits in 1964, followed by Intel’s 256-bit SRAM Intel 1101 in 1969, Intel’s 4-bit microprocessor 4004 in 1971, and so on. NMOS digital circuits were introduced around 1970. CMOS microprocessors were introduced in 1975 and overtook PMOS/NMOS processors before the 1980s.
- 18 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Solid-State Circuits Magazine
- Vol. 15 (3) , 36-42
Reflections on Progress in Digital Circuits: From basics to untangle past trials
A digital circuit is a system that discretizes and processes information and mainly expresses numerical values in binary numbers of 1 and 0. Logical 1 s and 0 s are physically represented by high and low voltages, large and small currents (or no current), and so on. Integrated digital circuits, called RTL , were introduced by Fairchild in 1961. To prevent short currents among input parallel registers, parallel diodes were introduced, called DTL . TTL, which replaced input diodes with bipolar transistors, were patented by James Buie of Pacific Semiconductor in 1961 and became the most popular logic configuration of the next two decades. General Microelectronics introduced the first commercial PMOS digital circuits in 1964, followed by Intel’s 256-bit SRAM Intel 1101 in 1969, Intel’s 4-bit microprocessor 4004 in 1971, and so on. NMOS digital circuits were introduced around 1970. CMOS microprocessors were introduced in 1975 and overtook PMOS/NMOS processors before the 1980s.
- 1 June 2023
- Published by Springer Nature
- p. 529-540
Lossless Method of Constraining Membrane Potential in Deep Spiking Neural Networks
- 1 June 2023
- Published by Springer Nature
- p. 529-540
Lossless Method of Constraining Membrane Potential in Deep Spiking Neural Networks