Pervesh Kumar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Pervesh Kumar has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
- Kang-Yoon Lee
Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, South Korea
6 shared publications - YoungGun Pu
Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, South Korea
6 shared publications - Imran Ali
Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon 16419, Korea
4 shared publications - Keum Cheol Hwang
Sungkyunkwan University
4 shared publications - Youngoo Yang
Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, South Korea
4 shared publications - Danial Khan
Innovative Technologies Laboratories (ITL), King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia
2 shared publications
Latest Publications
- Open Access
- 24 November 2022
- Published by MDPI AG in Electronics
- Vol. 11 (23) , 3883
A Study on the Design Procedure of Re-Configurable Convolutional Neural Network Engine for FPGA-Based Applications
Convolutional neural networks (CNNs) have become a primary approach in the field of artificial intelligence (AI), with wide range of applications. The two computational phases for every neural network are; the training phase and the testing phase. Usually, testing is performed on high-processing hardware engines, however, the training part is still a challenge for low-power devices. There are several neural accelerators; such as graphics processing units and field-programmable-gate-arrays (FPGAs). From the design perspective, an efficient hardware engine at the register-transfer level and efficient CNN modeling at the TensorFlow level are mandatory for any type of application. Hence, we propose a comprehensive, and step-by-step design procedure for a re-configurable CNN engine. We used TensorFlow and Keras libraries for modeling in Python, whereas the register-transfer-level part was performed using Verilog. The proposed idea was synthesized, placed, and routed for 180 nm complementary metal-oxide semiconductor technology using synopsis design compiler tools. The proposed design layout occupies an area of 3.16 × 3.16 mm2. A competitive accuracy of approximately 96% was achieved for the Modified National Institute of Standards and Technology (MNIST) and Canadian Institute for Advanced Research (CIFAR-10) datasets.
- 24 November 2022
- Published by MDPI AG in Electronics
- Vol. 11 (23) , 3883
Open AccessA Study on the Design Procedure of Re-Configurable Convolutional Neural Network Engine for FPGA-Based Applications
Convolutional neural networks (CNNs) have become a primary approach in the field of artificial intelligence (AI), with wide range of applications. The two computational phases for every neural network are; the training phase and the testing phase. Usually, testing is performed on high-processing hardware engines, however, the training part is still a challenge for low-power devices. There are several neural accelerators; such as graphics processing units and field-programmable-gate-arrays (FPGAs). From the design perspective, an efficient hardware engine at the register-transfer level and efficient CNN modeling at the TensorFlow level are mandatory for any type of application. Hence, we propose a comprehensive, and step-by-step design procedure for a re-configurable CNN engine. We used TensorFlow and Keras libraries for modeling in Python, whereas the register-transfer-level part was performed using Verilog. The proposed idea was synthesized, placed, and routed for 180 nm complementary metal-oxide semiconductor technology using synopsis design compiler tools. The proposed design layout occupies an area of 3.16 × 3.16 mm2. A competitive accuracy of approximately 96% was achieved for the Modified National Institute of Standards and Technology (MNIST) and Canadian Institute for Advanced Research (CIFAR-10) datasets.
- 8 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 38 (1) , 1348-1360
A Design of Wireless Power Receiver With Gate Charge Recycled Dual-Mode Active Rectifier and Step-Down Converter With 88.2% System Efficiency for Power Management IC
This article presents a power management integrated circuit for a wireless power receiver unit. A dual-mode high-efficiency active rectifier design is based on alliance for wireless power (A4WP) and wireless power consortium (WPC) standards. A gate charge recycling technique is proposed in the dual-mode active rectifier so that the current generated by the switching of the high side gate driver can be recycled to the rectifier output voltage in order to enhance efficiency. A step-down dc–dc converter with a proposed bootstrap circuit and dynamic pull-up resistor gate driver circuit is designed. The chip is implemented in a 0.18-μm bipolar-CMOS double-diffused metal-oxide-semiconductor process. The total die area, including the pad, is 2.64 mm × 2.6 mm. The measurement results show that the active rectifier achieves peak power conversion efficiency (PCE) of 94.7% and generates an output voltage of 7.68 V in WPC mode and 9.1 V with 93.4% efficiency in A4WPC mode. The input voltage varies from 10 V P-P –12 V P-P . Similarly, the step-down dc–dc converter obtains a peak PCE of 93.2% at a load current of 500 mA while producing an output voltage of 5 V. The overall system efficiency at A4WP and WPC mode are 86.7% and 88.2%, respectively.
- 8 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 38 (1) , 1348-1360
A Design of Wireless Power Receiver With Gate Charge Recycled Dual-Mode Active Rectifier and Step-Down Converter With 88.2% System Efficiency for Power Management IC
This article presents a power management integrated circuit for a wireless power receiver unit. A dual-mode high-efficiency active rectifier design is based on alliance for wireless power (A4WP) and wireless power consortium (WPC) standards. A gate charge recycling technique is proposed in the dual-mode active rectifier so that the current generated by the switching of the high side gate driver can be recycled to the rectifier output voltage in order to enhance efficiency. A step-down dc–dc converter with a proposed bootstrap circuit and dynamic pull-up resistor gate driver circuit is designed. The chip is implemented in a 0.18-μm bipolar-CMOS double-diffused metal-oxide-semiconductor process. The total die area, including the pad, is 2.64 mm × 2.6 mm. The measurement results show that the active rectifier achieves peak power conversion efficiency (PCE) of 94.7% and generates an output voltage of 7.68 V in WPC mode and 9.1 V with 93.4% efficiency in A4WPC mode. The input voltage varies from 10 V P-P –12 V P-P . Similarly, the step-down dc–dc converter obtains a peak PCE of 93.2% at a load current of 500 mA while producing an output voltage of 5 V. The overall system efficiency at A4WP and WPC mode are 86.7% and 88.2%, respectively.
- Open Access
A Low-Power Analog Processor-in-Memory-Based Convolutional Neural Network for Biosensor Applications
This paper presents an on-chip implementation of an analog processor-in-memory (PIM)-based convolutional neural network (CNN) in a biosensor. The operator was designed with low power to implement CNN as an on-chip device on the biosensor, which consists of plates of 32 × 32 material. In this paper, 10T SRAM-based analog PIM, which performs multiple and average (MAV) operations with multiplication and accumulation (MAC), is used as a filter to implement CNN at low power. PIM proceeds with MAV operations, with feature extraction as a filter, using an analog method. To prepare the input feature, an input matrix is formed by scanning a 32 × 32 biosensor based on a digital controller operating at 32 MHz frequency. Memory reuse techniques were applied to the analog SRAM filter, which is the core of low power implementation, and in order to accurately grasp the MAC operational efficiency and classification, we modeled and trained numerous input features based on biosignal data, confirming the classification. When the learned weight data was input, 19 mW of power was consumed during analog-based MAC operation. The implementation showed an energy efficiency of 5.38 TOPS/W and was differentiated through the implementation of 8 bits of high resolution in the 180 nm CMOS process.
Open AccessA Low-Power Analog Processor-in-Memory-Based Convolutional Neural Network for Biosensor Applications
This paper presents an on-chip implementation of an analog processor-in-memory (PIM)-based convolutional neural network (CNN) in a biosensor. The operator was designed with low power to implement CNN as an on-chip device on the biosensor, which consists of plates of 32 × 32 material. In this paper, 10T SRAM-based analog PIM, which performs multiple and average (MAV) operations with multiplication and accumulation (MAC), is used as a filter to implement CNN at low power. PIM proceeds with MAV operations, with feature extraction as a filter, using an analog method. To prepare the input feature, an input matrix is formed by scanning a 32 × 32 biosensor based on a digital controller operating at 32 MHz frequency. Memory reuse techniques were applied to the analog SRAM filter, which is the core of low power implementation, and in order to accurately grasp the MAC operational efficiency and classification, we modeled and trained numerous input features based on biosignal data, confirming the classification. When the learned weight data was input, 19 mW of power was consumed during analog-based MAC operation. The implementation showed an energy efficiency of 5.38 TOPS/W and was differentiated through the implementation of 8 bits of high resolution in the 180 nm CMOS process.
- Open Access
A Configurable and Fully Synthesizable RTL-Based Convolutional Neural Network for Biosensor Applications
This paper presents a register-transistor level (RTL) based convolutional neural network (CNN) for biosensor applications. Biosensor-based diseases detection by DNA identification using biosensors is currently needed. We proposed a synthesizable RTL-based CNN architecture for this purpose. The adopted technique of parallel computation of multiplication and accumulation (MAC) approach optimizes the hardware overhead by significantly reducing the arithmetic calculation and achieves instant results. While multiplier bank sharing throughout the convolutional operation with fully connected operation significantly reduces the implementation area. The CNN model is trained in MATLAB® on MNIST® handwritten dataset. For validation, the image pixel array from MNIST® handwritten dataset is applied on proposed RTL-based CNN architecture for biosensor applications in ModelSim®. The consistency is checked with multiple test samples and 92% accuracy is achieved. The proposed idea is implemented in 28 nm CMOS technology. It occupies 9.986 mm2 of the total area. The power requirement is 2.93 W from 1.8 V supply. The total time taken is 8.6538 ms.
Open AccessA Configurable and Fully Synthesizable RTL-Based Convolutional Neural Network for Biosensor Applications
This paper presents a register-transistor level (RTL) based convolutional neural network (CNN) for biosensor applications. Biosensor-based diseases detection by DNA identification using biosensors is currently needed. We proposed a synthesizable RTL-based CNN architecture for this purpose. The adopted technique of parallel computation of multiplication and accumulation (MAC) approach optimizes the hardware overhead by significantly reducing the arithmetic calculation and achieves instant results. While multiplier bank sharing throughout the convolutional operation with fully connected operation significantly reduces the implementation area. The CNN model is trained in MATLAB® on MNIST® handwritten dataset. For validation, the image pixel array from MNIST® handwritten dataset is applied on proposed RTL-based CNN architecture for biosensor applications in ModelSim®. The consistency is checked with multiple test samples and 92% accuracy is achieved. The proposed idea is implemented in 28 nm CMOS technology. It occupies 9.986 mm2 of the total area. The power requirement is 2.93 W from 1.8 V supply. The total time taken is 8.6538 ms.
- Open Access
- 26 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 152984-152992
An Ultra-Low-Power 2.4 GHz All-Digital Phase-Locked Loop With Injection-Locked Frequency Multiplier and Continuous Frequency Tracking
This paper presents a 0.46 mW and 2.4 GHz; All-Digital Phase-Locked Loop (ADPLL) through an Injection-Locked Frequency Multiplier (ILFM) and Continuous Frequency Tracking Loop (CFTL) circuitry for low power Internet-of-Thing (IoT) applications. In the proposed ADPLL architecture to save power, the need for Time-to-Digital Converter (TDC) is eliminated through providing the CFTL circuitry. This feature makes the design compact, low power, and suitable for IoT applications. The proposed design is based on a synthesizable pulse injection and frequency-locked loop along with an ultra-low-power LC Digitally-Controlled Oscillator (LC-DCO). The presented CFTL circuit adjusts the frequency of the DCO continuously and prevents the frequency drift after the reference injection. Inside the designed LC-DCO core, the power consumption is minimized by optimizing the $\text{g}_{\mathrm {m}}~/\text{I}_{\mathrm {D}}$ and adjusting the power supply to 0.5 V. The proposed ILFM based ADPLL is fabricated in 55 nm CMOS technology and covers the operational frequency range of 2.402 GHz to 2.480 GHz with a reference frequency of 32 MHz. The measured phase noise performance of the ADPLL is −111.15 dBc/Hz at 1 MHz offset frequency from the carrier frequency of 2.4 GHz. It consumes only 0.46 mW power with an active area of 0.129 mm 2 .
- 26 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 152984-152992
Open AccessAn Ultra-Low-Power 2.4 GHz All-Digital Phase-Locked Loop With Injection-Locked Frequency Multiplier and Continuous Frequency Tracking
This paper presents a 0.46 mW and 2.4 GHz; All-Digital Phase-Locked Loop (ADPLL) through an Injection-Locked Frequency Multiplier (ILFM) and Continuous Frequency Tracking Loop (CFTL) circuitry for low power Internet-of-Thing (IoT) applications. In the proposed ADPLL architecture to save power, the need for Time-to-Digital Converter (TDC) is eliminated through providing the CFTL circuitry. This feature makes the design compact, low power, and suitable for IoT applications. The proposed design is based on a synthesizable pulse injection and frequency-locked loop along with an ultra-low-power LC Digitally-Controlled Oscillator (LC-DCO). The presented CFTL circuit adjusts the frequency of the DCO continuously and prevents the frequency drift after the reference injection. Inside the designed LC-DCO core, the power consumption is minimized by optimizing the $\text{g}_{\mathrm {m}}~/\text{I}_{\mathrm {D}}$ and adjusting the power supply to 0.5 V. The proposed ILFM based ADPLL is fabricated in 55 nm CMOS technology and covers the operational frequency range of 2.402 GHz to 2.480 GHz with a reference frequency of 32 MHz. The measured phase noise performance of the ADPLL is −111.15 dBc/Hz at 1 MHz offset frequency from the carrier frequency of 2.4 GHz. It consumes only 0.46 mW power with an active area of 0.129 mm 2 .
- Open Access
A Design of Adaptive Control and Communication Protocol for SWIPT System in 180 nm CMOS Process for Sensor Applications
This paper presents an adaptive control and communication protocol (ACCP) for the ultra-low power simultaneous wireless information and power transfer (SWIPT) system for sensor applications. The SWIPT system-related operations depend on harvested radio frequency (RF) energy from the ambient environment. The necessary power for SWIPT system operation is not always available and it depends on the available RF energy in the ambient environment, transmitted RF power from the SWIPT transmitter, and the distance from the transmitter and channel conditions. Thus, an efficient control and communication protocol is required which can control the SWIPT system for sensor applications which mainly consists of a transmitter and a receiver. Multiple data frame structures are used to optimize the exchange of bits for the communication and control of the SWIPT system. Both SWIPT transmitter and receiver are capable of using multiple modulation schemes which can be switched depending on the channel condition and the available RF energy in the ambient environment. This provides support for automatic switching between the time switching scheme and power splitting scheme for the distribution of received RF power in the SWIPT receiver. It also adjusts the digital clock frequency at the SWIPT receiver according to the harvested power level to optimize the power consumption. The SWIPT receiver controller with ACCP is implemented in 180 nm CMOS technology. The RF frequency of the SWIPT operation is 5.8 GHz. Digital clock frequency at the SWIPT receiver can be adjusted between 32 kHz and 2 MHz which provides data rates from 8 to 500 kbps, respectively. The power consumption and area utilization are 12.3 µW and 0.81 mm².
Open AccessA Design of Adaptive Control and Communication Protocol for SWIPT System in 180 nm CMOS Process for Sensor Applications
This paper presents an adaptive control and communication protocol (ACCP) for the ultra-low power simultaneous wireless information and power transfer (SWIPT) system for sensor applications. The SWIPT system-related operations depend on harvested radio frequency (RF) energy from the ambient environment. The necessary power for SWIPT system operation is not always available and it depends on the available RF energy in the ambient environment, transmitted RF power from the SWIPT transmitter, and the distance from the transmitter and channel conditions. Thus, an efficient control and communication protocol is required which can control the SWIPT system for sensor applications which mainly consists of a transmitter and a receiver. Multiple data frame structures are used to optimize the exchange of bits for the communication and control of the SWIPT system. Both SWIPT transmitter and receiver are capable of using multiple modulation schemes which can be switched depending on the channel condition and the available RF energy in the ambient environment. This provides support for automatic switching between the time switching scheme and power splitting scheme for the distribution of received RF power in the SWIPT receiver. It also adjusts the digital clock frequency at the SWIPT receiver according to the harvested power level to optimize the power consumption. The SWIPT receiver controller with ACCP is implemented in 180 nm CMOS technology. The RF frequency of the SWIPT operation is 5.8 GHz. Digital clock frequency at the SWIPT receiver can be adjusted between 32 kHz and 2 MHz which provides data rates from 8 to 500 kbps, respectively. The power consumption and area utilization are 12.3 µW and 0.81 mm².