• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.9
• /
• pp.1158-1165
• /
• 2021

This paper proposes a CNN accelerator which is optimized Pooling layer operation incorporated in Multiplication And Accumulation(MAC) to reduce the memory size. For optimizing memory and data path circuit, the quantized 8bit integer weights are used instead of 32bit floating-point weights for pre-training of MNIST data set. To reduce chip area, the proposed CNN model is reduced by a convolutional layer, a 4*4 Max Pooling, and two fully connected layers. And all the operations use specific MAC with approximation adders and multipliers. 94% of internal memory size reduction is achieved by simultaneously performing the convolution and the pooling operation in the proposed architecture. The proposed accelerator chip is designed by using TSMC65nmGP CMOS process. That has about half size of our previous paper, 0.8*0.9 = 0.72mm². The presented CNN accelerator chip achieves 94% accuracy and 77us inference time per an MNIST image.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.14 no.6
• /
• pp.479-486
• /
• 2021

In this paper, an ultra small phase locked loop (PLL) with a single capacitor loop filter has been proposed by adding a signal sensing circuit (SSC). In order to extremely reduce the size of the PLL, the passive element loop filter, which occupies the largest area, is designed with a very small single capacitor (2pF). The proposed PLL is designed to operate stably by the output of the internal negative feedback loop including the SSC acting as a negative feedback to the output of the single capacitor loop filter of the external negative feedback loop. The SSC that detects the PLL output signal change reduces the excess phase shift of the PLL output frequency by adjusting the capacitance charge of the loop filter. Although the proposed structure has a capacitor that is 1/78 smaller than that of the existing structure, the jitter size differs by about 10%. The PLL is designed using a 1.8V 180nm CMOS process and the Spice simulation results show that it works stably.

• International journal of advanced smart convergence
• /
• v.11 no.4
• /
• pp.96-103
• /
• 2022

This paper presents the analysis of increasing the resolution of True-Time-Delay (TTD) by 0.5-bit for phased-array antenna system which is one of the Multiple-Input and Multiple Output (MIMO) technologies. For the analysis, a 5.5-bit True-Time Delay (TTD) integrated circuit is designed and analyzed in terms of beam steering performance. In order to increase the number of effective bits, the designed 5.5-bit TTD uses Single Pole Triple Throw (SP3T) and Double Pole Triple Throw (DP3T) switches, and this method can minimize the circuit area by inserting the minimum time delay of 0.5-bit. Furthermore, the circuit mostly maintains the performance of the circuit with the fully added bits. The idea of adding 0.5-bit is verified by analyzing the relation between the number of bits and array elements. The 5.5-bit TTD is designed using 0.18 ㎛ RF CMOS process and the estimated size of the designed circuit excluding the pad is 0.57×1.53 mm². In contrast to the conventional phase shifter which has distortion of scanning angle known as beam squint phenomenon, the proposed TTD circuit has constant time delays for all states across a wide frequency range of 4 - 20 GHz with minimized power consumption. The minimum time delay is designed to have 1.1 ps and 2.2 ps for the 0.5-bit option and the normal 1-bit option, respectively. A simulation for beam patterns where the 10 phased-array antenna is assumed at 10 GHz confirms that the 0.5-bit concept suppresses the pointing error and the relative power error by up to 1.5 degrees and 80 mW, respectively, compared to the conventional 5-bit TTD circuit.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.9
• /
• pp.58-67
• /
• 2009

This paper presents a fully digital gain control system with a new high bandwidth and wide dynamic range power detector for DVB-S2 application. Because the peak-to-average power ratio (PAPR) of DVB-S2 system is so high and the settling time requirement is so stringent, the conventional closed-loop analog gain control scheme cannot be used. The digital gain control is necessary for the robust gain control and the direct digital interface with the baseband modem. Also, it has several advantages over the analog gain control in terms of the settling time and insensitivity to the process, voltage and temperature variation. In order to have a wide gain range with fine step resolution, a new AGC system is proposed. The system is composed of high-bandwidth digital VGAs, wide dynamic range power detectors with RMS detector, low power SAR type ADC, and a digital gain controller. To reduce the power consumption and chip area, only one SAR type ADC is used, and its input is time-interleaved based on four power detectors. Simulation and measurement results show that the new AGC system converges with gain error less than 0.25 dB to the desired level within $10{\mu}s$. It is implemented in a $0.18{\mu}m$ CMOS process. The measurement results of the proposed IF AGC system exhibit 80-dB gain range with 0.25-dB resolution, 8 nV/$\sqrt{Hz}$ input referred noise, and 5-dBm $IIP_3$ at 60-mW power consumption. The power detector shows the 35dB dynamic range for 100 MHz input.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2008.11a
• /
• pp.10-10
• /
• 2008

Silicide is inevitable for CMOSFETs to reduce RC delay by reducing the sheet resistance of gate and source/drain regions. Ni-silicide is a promising material which can be used for the 65nm CMOS technologies. Ni-silicide was proposed in order to make up for the weak points of Co-silicide and Ti-silicide, such as the high consumption of silicon and the line width limitation. Low resistivity NiSi can be formed at low temperature ($\sim500^{\circ}C$) with only one-step heat treat. Ni silicide also has less dependence of sheet resistance on line width and less consumption of silicon because of low resistivity NiSi phase. However, the low thermal stability of the Ni-silicide is a major problem for the post process implementation, such as metalization or ILD(inter layer dielectric) process, that is, it is crucial to prevent both the agglomeration of mono-silicide and its transformation into $NiSi_2$. To solve the thermal immune problem of Ni-silicide, various studies, such as capping layer and inter layer, have been worked. In this paper, the Ni-silicide utilizing Pd stacked layer (Pd/Ni/TiN) was studied for highly thermal immune nano-scale CMOSFETs technology. The proposed structure was compared with NiITiN structure and showed much better thermal stability than Ni/TiN.

• Korean Journal of Materials Research
• /
• v.10 no.12
• /
• pp.812-818
• /
• 2000

We studied the effects of Ge preamorphization (PAM) on 0.25$\mu\textrm{m}$ Ti-salicide junctions using comparative study with As PAM. For each PAM schemes, ion implantations are performed at a dose of 2E14 ion/$\textrm{cm}^2$ and at 20keV energy using $^{75}$ /As+and GeF4 ion sources. Ge PAM showed better sheet resistance and within- wafer uniformity than those of As PAM at 0.257m line width of n +/p-well junctions. This attributes to enhanced C54-silicidation reaction and strong (040) preferred orientation of the C54-silicide due to minimized As presence at n+ junctions. At p+ junctions, comparable performance was obtained in Rs reduction at fine lines from both As and Ge PAM schemes. Junction leakage current (JLC) revels are below ~1E-14 A/$\mu\textrm{m}^{2}$ at area patterns for all process conditions, whereas no degradation in JLC is shown under Ge PAM condition even at edge- intensive patterns. Smooth $TiSi_2$ interface is observed by cross- section TEM (X- TEM), which supports minimized silicide agglomeration due to Ge PAM and low level of JLC. Both junction break- down voltage (JBV) and contact resistances are satisfactory at all process conditions.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.9C
• /
• pp.861-870
• /
• 2002

In this paper, with targeting on the drawback of RSA of operation speed, a new 1024-bit RSA cryptosystem has been proposed and implemented in hardware to increase the operational speed and perform the variable-length encryption. The proposed cryptosystem mainly consists of the modular exponentiation part and the modular multiplication part. For the modular exponentiation, the RL-binary method, which performs squaring and modular multiplying in parallel, was improved, and then applied. And 4-stage CSA structure and radix-4 booth algorithm were applied to enhance the variable-length operation and reduce the number of partial product in modular multiplication arithmetic. The proposed RSA cryptosystem which can calculate at most 1024 bits at a tittle was mapped into the integrated circuit using the Hynix Phantom Cell Library for Hynix 0.35㎛ 2-Poly 4-Metal CMOS process. Also, the result of software implementation, which had been programmed prior to the hardware research, has been used to verify the operation of the hardware system. The size of the result from the hardware implementation was about 190k gate count and the operational clock frequency was 150㎒. By considering a variable-length of modulus number, the baud rate of the proposed scheme is one and half times faster than the previous works. Therefore, the proposed high speed variable-length RSA cryptosystem should be able to be used in various information security system which requires high speed operation.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.44 no.6 s.360
• /
• pp.19-27
• /
• 2007

This paper presents power supply-insensitive Gbps low power LVDS I/O circuits. The proposed LVDS I/O has been designed and simulated using 1.8V, $0.18\;{\mu}m$ TSMC CMOS Process. The LVDS I/O includes transmitter and receiver parts. The transmitter circuits consist of a differential phase splitter and an output stage with the switched capacitor common mode feedback(SC-CMFB). The differential phase splitter generates a pair of differential signals which provides a balanced duty $cycle(50{\pm}2%)$ and phase difference$(180{\pm}0.2^{\circ})$ over a wide supply voltage range. Also, $V_{OD}$ voltage is 250 mV which is the smallest value of the permissible $V_{OD}$ range for low power operation. The output buffer maintains the required $V_{CM}$ within the permissible range$(1.2{\pm}0.1V)$ due to the SC-CMFB. The receiver covers a wide input DC offset $range(0.2{\sim}2.6\;V)$ with 38 mV hysteresis and Produces a rail-to-rail output over a wide supply voltage range. Beside, the designed receiver has 38.9 dB gain at 1 GHz, which is higher than conventional receivers.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.8
• /
• pp.95-101
• /
• 2009

Random numbers are used in many sorts of applications. Some applications, like simple software simulation tests, communication protocol verifications, cryptography verification and so forth, need various levels of randomness with various process speeds. In this paper, we propose a fast pseudorandom generator module for embedded systems. The generator module is implemented in hardware which can run in two modes, one of which can generate random numbers with higher randomness but which requires six cycles, the other providing its result within one cycle but with less randomness. An ASIP (Application Specific Instruction set Processor) was designed to implement the proposed pseudorandom generator instruction sets. We designed a processor based on the MIPS architecture,, by using LISA, and have run statistical tests passing the sequence of the Diehard test suite. The HDL models of the processor were generated using CoWare's Processor Designer and synthesized into the Dong-bu 0.18um CMOS cell library using the Synopsys Design Compiler. With the proposed pseudorandom generator module, random number generation performance was 239% faster than software model, but the area increased only 2.0% of the proposed ASIP.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.8
• /
• pp.38-48
• /
• 2014

In this paper, a design of Signal Processing Analog Front-End IC for Automotive Piezo-Resistive Type Pressure Sensor is presented. In modern society, as the car turns to go from mechanical to electronic technology, the accuracy and reliability of electronic parts required importantly. In order to improve these points, Programmable Gain Amplifier (PGA) amplifies the received signal in accordance with gain for increasing the accuracy after PRT Sensor is operated to change physical pressure signals to electrical signals. The signal amplified from PGA is processed by Digital blocks like ADC, CMC and DAC. After going through this process, it is possible to determine the electrical signal to physical pressure signal. As processing analog signal to digital signal, reliability and accuracy in Analog Front-End IC is increased. The current consumption of IC is 5.32mA. The die area of the fabricated IC is $1.94mm{\times}1.94mm$.