Sara Choi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sara Choi has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Seong-Ook Jung
Yonsei University
5 shared publications - Seung H. Kang
Qualcomm Technologies, Inc., San Diego, CA, USA
4 shared publications Byungkyu Song
Department of Electrical and Electronic Engineering, Yonsei University, Seoul 03722, South Korea
3 shared publications- Hong Keun Ahn
School of Electrical and Electronic Engineering, Yonsei University, Seoul, Republic of Korea
2 shared publications Jung Pill Kim
Qualcomm Incorporated, San Diego, CA 92121 USA
2 shared publicationsByung Kyu Song
School of Electrical and Electronic Engineering, Yonsei University, Seoul 120-749, South Korea
1 shared publication
Latest Publications
- 7 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 68 (6) , 2481-2493
Self-Referenced Single-Ended Resistance Monitoring Write Termination Scheme for STT-RAM Write Energy Reduction
Essential design requirements for a sense amplifier (SA) used in the resistance monitoring write termination (RM-WT) scheme are suggested to reduce the write energy of spin-transfer-torque random access memory (STT-RAM) while achieving a write pass yield comparable to that of a conventional write operation. In addition, a self-referenced single-ended RM-WT (SS-RM-WT) scheme is proposed. To reduce the offset voltage, a single-ended sensing circuit (SE-SC) is used in the SA. A data-aware input voltage-transfer method is also adopted in the SE-SC to maximize the input voltage difference. By adopting a capacitor between the output of the SE-SC and the input of an inverter generating a logical output used for the write termination, the conflict between maintaining and changing the output of the SE-SC is resolved. The simulation results using the industry-compatible 65-nm technology HSPICE model parameters show that the proposed SS-RM-WT scheme achieves a 44% write energy saving on average without increasing the write error rate. Area overhead is only 11.8% for a 256-kb STT-RAM array, whereas that of the previous self-referenced RM-WT schemes is up to 42.5%.
- 7 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 68 (6) , 2481-2493
Self-Referenced Single-Ended Resistance Monitoring Write Termination Scheme for STT-RAM Write Energy Reduction
Essential design requirements for a sense amplifier (SA) used in the resistance monitoring write termination (RM-WT) scheme are suggested to reduce the write energy of spin-transfer-torque random access memory (STT-RAM) while achieving a write pass yield comparable to that of a conventional write operation. In addition, a self-referenced single-ended RM-WT (SS-RM-WT) scheme is proposed. To reduce the offset voltage, a single-ended sensing circuit (SE-SC) is used in the SA. A data-aware input voltage-transfer method is also adopted in the SE-SC to maximize the input voltage difference. By adopting a capacitor between the output of the SE-SC and the input of an inverter generating a logical output used for the write termination, the conflict between maintaining and changing the output of the SE-SC is resolved. The simulation results using the industry-compatible 65-nm technology HSPICE model parameters show that the proposed SS-RM-WT scheme achieves a 44% write energy saving on average without increasing the write error rate. Area overhead is only 11.8% for a 256-kb STT-RAM array, whereas that of the previous self-referenced RM-WT schemes is up to 42.5%.
- Open Access
- 2 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 216922-216932
Area- and Energy-Efficient STDP Learning Algorithm for Spiking Neural Network SoC
Recently, spiking neural networks have gained attention owing to their energy efficiency. All-to-all spike-time dependent plasticity is a popular learning algorithm for spiking neural networks because it is suitable for nondifferentiable spike event-based learning and requires fewer computations than back-propagation-based algorithms. However, the hardware implementation of all-to-all spike-time dependent plasticity is limited by the large storage area required for spike history and large energy consumption caused by frequent memory access. We propose a time-step scaled spike-time dependent plasticity to reduce the storage area required for spike history by reducing the area of the spike-time dependent plasticity learning circuit by 60% and a post-neuron spike-referred spike-time dependent plasticity to reduce the energy consumption by 99.1% by efficiently accessing the memory while learning. The accuracy of Modified National Institute of Standards and Technology image classification degraded by less than 2% when both time-step scaled spike-time dependent plasticity and post-neuron spike-referred spike-time dependent plasticity were applied. Thus, the proposed hardware-friendly spike-time dependent plasticity algorithms make all-to-all spike-time dependent plasticity implementable in more compact areas while reducing energy consumption and experiencing insignificant accuracy degradation.
- 2 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 216922-216932
Open AccessArea- and Energy-Efficient STDP Learning Algorithm for Spiking Neural Network SoC
Recently, spiking neural networks have gained attention owing to their energy efficiency. All-to-all spike-time dependent plasticity is a popular learning algorithm for spiking neural networks because it is suitable for nondifferentiable spike event-based learning and requires fewer computations than back-propagation-based algorithms. However, the hardware implementation of all-to-all spike-time dependent plasticity is limited by the large storage area required for spike history and large energy consumption caused by frequent memory access. We propose a time-step scaled spike-time dependent plasticity to reduce the storage area required for spike history by reducing the area of the spike-time dependent plasticity learning circuit by 60% and a post-neuron spike-referred spike-time dependent plasticity to reduce the energy consumption by 99.1% by efficiently accessing the memory while learning. The accuracy of Modified National Institute of Standards and Technology image classification degraded by less than 2% when both time-step scaled spike-time dependent plasticity and post-neuron spike-referred spike-time dependent plasticity were applied. Thus, the proposed hardware-friendly spike-time dependent plasticity algorithms make all-to-all spike-time dependent plasticity implementable in more compact areas while reducing energy consumption and experiencing insignificant accuracy degradation.
- 18 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 27 (11) , 2548-2555
Offset-Canceling Single-Ended Sensing Scheme With One-Bit-Line Precharge Architecture for Resistive Nonvolatile Memory in 65-nm CMOS
In the design of nonvolatile memory (NVM), the sensing scheme (SS) has become a read-energy bottleneck because the required read-cell current is too large to satisfy a target read yield. This problem is further aggravated by technology scaling because increased process variation and reduced supply voltage (VDD) require more current to satisfy the target read yield. This paper proposes an offset-canceling single-ended SS (OCSE-SS) with one-bit-line precharge architecture (1BLPA) that is intended for use in ultralow power NVM applications. The test chip is fabricated using 65-nm process technology, and the measurement results show that the read energy per bit of the OCSE-SS is 1/3 compared to that of the conventional SS (Conv-SS). The read energy reduction comes from the single-ended sensing, offset cancellation, and 1BLPA features. Moreover, when a resistance difference between the data and reference cells is as small as 0.5 kΩ , the OCSE-SS reads successfully with a VDD of 1.0 V and a sensing time (tSEN) of 17 ns due to the offset cancellation characteristic, whereas the Conv-SS fails regardless of VDD and tSEN values.
- 18 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 27 (11) , 2548-2555
Offset-Canceling Single-Ended Sensing Scheme With One-Bit-Line Precharge Architecture for Resistive Nonvolatile Memory in 65-nm CMOS
In the design of nonvolatile memory (NVM), the sensing scheme (SS) has become a read-energy bottleneck because the required read-cell current is too large to satisfy a target read yield. This problem is further aggravated by technology scaling because increased process variation and reduced supply voltage (VDD) require more current to satisfy the target read yield. This paper proposes an offset-canceling single-ended SS (OCSE-SS) with one-bit-line precharge architecture (1BLPA) that is intended for use in ultralow power NVM applications. The test chip is fabricated using 65-nm process technology, and the measurement results show that the read energy per bit of the OCSE-SS is 1/3 compared to that of the conventional SS (Conv-SS). The read energy reduction comes from the single-ended sensing, offset cancellation, and 1BLPA features. Moreover, when a resistance difference between the data and reference cells is as small as 0.5 kΩ , the OCSE-SS reads successfully with a VDD of 1.0 V and a sensing time (tSEN) of 17 ns due to the offset cancellation characteristic, whereas the Conv-SS fails regardless of VDD and tSEN values.
- 16 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 66 (8) , 2963-2972
Offset-Cancellation Sensing-Circuit-Based Nonvolatile Flip-Flop Operating in Near-Threshold Voltage Region
For ultra-low-power Internet of Things (IoT) devices powered by batteries or energy harvesting, a nonvolatile flip-flop (NV-FF) operating in the near-threshold voltage (NTV) region has been widely studied to save both standby and active power. However, in the NTV region, the NV-FF suffers from restore yield and performance degradation due to the increase in process variation combined with the decrease in supply voltage. This paper presents a comparative analysis of the previous NV-FFs according to supply voltage scaling and suggests the design directions for implementing the NV-FF that can operate with compact area and high performance even in the NTV region. Through the comparative analysis, a novel offset-cancellation sensing-circuit (OCSC)-based NV-FF, which adopts a separate latch and sensing circuit structure and an offset-cancellation technique, is proposed. The Monte Carlo HSPICE simulation results using industry-compatible 65-nm model parameters show that the proposed NV-FF satisfies a target restore yield with 32% area, 88% speed, and 82% energy savings in comparison with a representative NV-FF of the merged latch and sensing circuit structure.
- 16 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 66 (8) , 2963-2972
Offset-Cancellation Sensing-Circuit-Based Nonvolatile Flip-Flop Operating in Near-Threshold Voltage Region
For ultra-low-power Internet of Things (IoT) devices powered by batteries or energy harvesting, a nonvolatile flip-flop (NV-FF) operating in the near-threshold voltage (NTV) region has been widely studied to save both standby and active power. However, in the NTV region, the NV-FF suffers from restore yield and performance degradation due to the increase in process variation combined with the decrease in supply voltage. This paper presents a comparative analysis of the previous NV-FFs according to supply voltage scaling and suggests the design directions for implementing the NV-FF that can operate with compact area and high performance even in the NTV region. Through the comparative analysis, a novel offset-cancellation sensing-circuit (OCSC)-based NV-FF, which adopts a separate latch and sensing circuit structure and an offset-cancellation technique, is proposed. The Monte Carlo HSPICE simulation results using industry-compatible 65-nm model parameters show that the proposed NV-FF satisfies a target restore yield with 32% area, 88% speed, and 82% energy savings in comparison with a representative NV-FF of the merged latch and sensing circuit structure.
- 27 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 27 (2) , 387-397
A Decoder for Short BCH Codes With High Decoding Efficiency and Low Power for Emerging Memories
In this paper, a double-error-correcting and triple-error-detecting (DEC-TED) Bose-Chaudhuri-Hocquenghem (BCH) code decoder with high decoding efficiency and low power for error correction in emerging memories is presented. To increase the decoding efficiency, we propose an adaptive error correction technique for the DEC-TED BCH code that detects the number of errors in a codeword immediately after syndrome generation and applies a different error correction algorithm depending on the error conditions. With the adaptive error correction technique, the average decoding latency and power consumption are significantly reduced owing to the increased decoding efficiency. To further reduce the power consumption, an invalid-transition-inhibition technique is proposed to remove the invalid transitions caused by glitches of syndrome vectors in the error-finding block. Synthesis results with an industry-compatible 65-nm technology library show that the proposed decoders for the (79, 64, 6) BCH code take only 37%-48% average decoding latency and achieve more than 70% power reduction compared to the conventional fully parallel decoder under the 10⁻⁴-10⁻² raw bit-error rate.
- 27 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 27 (2) , 387-397
A Decoder for Short BCH Codes With High Decoding Efficiency and Low Power for Emerging Memories
In this paper, a double-error-correcting and triple-error-detecting (DEC-TED) Bose-Chaudhuri-Hocquenghem (BCH) code decoder with high decoding efficiency and low power for error correction in emerging memories is presented. To increase the decoding efficiency, we propose an adaptive error correction technique for the DEC-TED BCH code that detects the number of errors in a codeword immediately after syndrome generation and applies a different error correction algorithm depending on the error conditions. With the adaptive error correction technique, the average decoding latency and power consumption are significantly reduced owing to the increased decoding efficiency. To further reduce the power consumption, an invalid-transition-inhibition technique is proposed to remove the invalid transitions caused by glitches of syndrome vectors in the error-finding block. Synthesis results with an industry-compatible 65-nm technology library show that the proposed decoders for the (79, 64, 6) BCH code take only 37%-48% average decoding latency and achieve more than 70% power reduction compared to the conventional fully parallel decoder under the 10⁻⁴-10⁻² raw bit-error rate.