• Proceedings of the KIEE Conference
• /
• 1993.07a
• /
• pp.254-257
• /
• 1993

This paper presents a new method to implement Hebbian learning method on artificial neural network. In hebbian learning algorithm, complexity in terms of multiplications is high. To save the chip area, we consider a new learning circuit. By calculating similarity, or correlation between $X_i$ and $O_i$, large portion of circuits commonly used in conventional neural networks is not necessary for this new hebbian learning circuit named COR. The output signals of COR is applied to weight storage capacitors for direct control the voltages of the capacitors. The weighted sum, ${\Sigma}W_{ij}O_j$, is realized by multipliers, whose output currents are summed up in one line which goes to learning circuit or output circuit. The drain current of the multiplier can produce positive or negative synaptic weights. The pass transistor selects eight learning mode or recall mode. The layout of an learnable six-neuron fully connected Hopfield neural network is designed, and is simulated using PSPICE. The network memorizes, and retrieves the patterns correctly under the existence of minor noises.

• The Journal of the Korea Contents Association
• /
• v.21 no.7
• /
• pp.164-170
• /
• 2021

In VLSI design, partitioning is a task of clustering objects into groups so that a given objective circuit is optimized. It is used at the layout level to find strongly connected components that can be placed together in order to minimize the layout area and propagation delay. The most popular algorithms for partitioning include the Kernighan-Lin algorithm, Fiduccia-Mattheyses heuristic and simulated annealing. In this paper, we propose a adapted genetic algorithm searching solution space for the circuit partitioning problem, and then compare it with simulated annealing and genetic algorithm by analyzing the results of implementation. As a result, it was found that an adaptive genetic algorithm approaches the optimal solution more effectively than the simulated annealing and genetic algorithm.

• Proceedings of the IEEK Conference
• /
• 2002.07a
• /
• pp.650-653
• /
• 2002

The increaseing speeds are accompanied by decreases in pulse rise and fall time in VLSI circuits. These accenturate the high frequency spectral contents of the signals and cause the frequency dependent loss of the conductors which interconnect the various sub-circuits composing of VLSI circuit. The lossy effect is approximated by the square root of frequency dependence of the per unit length resistance. In the practical applications, several problems may arise along with this approximation, so we extend our investigation of the lossy effect by numerical Laplace inversion method.

• Proceedings of the IEEK Conference
• /
• 1998.10a
• /
• pp.1057-1060
• /
• 1998

Power Dissipation and circuit speed become the most importance parameters in VLSI system maximum power dissipation for VLSI system design. We remodeled CMOS inverter according to the operating region, saturation region or linear regin, and calculate maximum power dissipation point of CMOS inverter. The result of proposed maximum power dissipation model compared with those from SPICE simulation which results that the proposed maximum power dissipation model has the error rate within 10% to SPICE simulation.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2002.11b
• /
• pp.545-548
• /
• 2002

본 논문에서는 유리 B 스플라인 곡선을 이용한 블록 효과 감소 알고리듬을 VHDL을 이용하여 설계하고, 모의 실험을 통하여 동작을 확인한다. 블록 효과는 매우 낮은 비트율로 블록 기반 부호화 방식을 수행할 때 복원 영상에서 나타나는 블록 형태의 왜곡을 의미한다. 설계된 회로는 유리 B 스플라인 곡선을 적용한 블록 효과 감소 알고리듬으로서, 이 기법은 컴퓨터 그래픽스 분야에서 제어점을 근사하는 부드러운 곡선을 생성하기 위해 사용되는 스플라인 곡선을 적용하여 블록 현상을 효과적으로 감소시킨다. 설계된 회로는 주파수 100MHz에서 동작을 시켰으며, 모의 실험 결과 매우 우수한 블록 효과 감소 기능을 가진 것을 알 수 있다.

• Proceedings of the IEEK Conference
• /
• 1999.11a
• /
• pp.863-866
• /
• 1999

In this paper, we designed a Viterbi decoder with constraint length K=7, code rate R=1/2, encoder generator polynomial (171, 133)$_{8}$. This decoder makes use of 3-bit soft decision. We designed the Viterbi decoder using VHDL. We employed conventional logic circuit instead of ROM for branch metric units(BMUs) to reduce the number of gates. We adopted fully parallel structures for add-compare-select units(ACSUs). The size of the designed decoder is about 200, 000 gates.s.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.41 no.4
• /
• pp.101-112
• /
• 2004

In this paper, we developed a numerical analysis model by using ADI-FDTD method to analyze three-dimensional interconnect structure. We discretized maxwell's curl equation by using ADI-FDTD. Using ADI-FDTD method, a sampler circuit designed from 3.3 V CMOS technology is simplified to 3-metal line structure. Using this simplified structure, the time delay and signal distortion of complex interconnects are investigated. As results of simulation, 5∼10 ps of delay time and 0.1∼0.2 V of signal distortion are measured. As demonstrated in this paper, the full-wave analysis using ADI-FDTD exhibits a promise for accurate modeling of electromagnetic phenomena in high-speed VLSI interconnect.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.27 no.11
• /
• pp.126-134
• /
• 1990

In order to implement high-speed data acquistion system in CMOS VLSI technology, means must be found to overcome the relatively low transconductance and large device mismatch characteristic of MOS device. Because of these device limitations, circuit design approaches tradition-ally used in high-speed bipolar analog-to-digital converter(ADC) are suited to CMOS implementation. Also the design of VLSI CMOS comparator wherein voltage comparision is accomplished by means of a pipelined cascade RSA (Regenerative Sense Amplifier). So, in this paper we designed the A/D converter incorporates the pipelined CMOS comparator.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.12A
• /
• pp.2059-2066
• /
• 2001

In this paper, we proposed a method to reduce the number of input pins in one-dimensional VLSI array for fractal image compression. We use quad-tree partition scheme and can reduce the number of the input pins up to 50% by sharing the domain\`s and the range\`s data input pins in the proposed VLSI array architecture. Also, we can reduce the input pins and simplify the internal operation circuit of the processing elements by eliminating a few number of bits of the least significant bits of the input data. We simulated using the 256$\times$256 and 512$\times$512 Lena images to verify performance of the proposed method. As the result of simulation, we can decompress the original image with about 32dB(PSNR) in spite of elimination of the least significant 2-bit in the original input data, and additionally reduce the number of input pins up to 25% compared to VLSI array sharing input pins of range and domain.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.36C no.5
• /
• pp.13-26
• /
• 1999

This paper presents a metric that can guide to optimal circuit models for interconnects among various models, given interconnect parameters and operating environment. To get this goal, we categorize interconnects into RC~c1ass and RLC-c1ass model domains based on the quantitative modeling error analysis using total resistance, inductance and capacitance of interconnects as well as operating frequency. RC~c1ass circuit models, which include most on~chip interconnects, can be efficiently analyzed by using the model~order reduction techniques. RLC-c1ass circuit models are constructed using one of three candidates, ILC(Iterative Ladder Circuit) macromodels, MC(Method of Characteristics) macromodels, and state-based convolution method, the selection process of which is based upon the allowable modeling error and electrical parameters of interconnects. We propose the model domain diagram leading to optimal circuit models and the division of model domains has been achieved considering the simulation cost of macromodels under the environmental assumption of the general purpose circuit simulator such as SPICE. The macromodeling method presented in this paper keeps the passivity of the original interconnects and accordingly guarantees the unconditional stability of circuit models.