• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 1986.10a
• /
• pp.106-109
• /
• 1986

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 1985.10a
• /
• pp.220-223
• /
• 1985

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.15 no.2
• /
• pp.155-165
• /
• 1990

The problem of quantization errors in digital filter design arises because of the practical necessity due to finite wordlength implementation. These errors are classified into coefficient quantization error and roung off error. In this paper, in order to analyze and reduce these errors, minimum ceefficient quantization realization is directly derived form impulse responese design specification. And using the equivalent transform relation between minimum coefficient quantization error and minimum roundoff error realizations, we synthesize an optimal realization state-space digital filter. This technique is analyzed by the simulation of an approximated 3rd model, which shows that it is superior to direct or cascade state-space digital filter in quantization errors.

• ETRI Journal
• /
• v.45 no.1
• /
• pp.119-130
• /
• 2023

We propose a quantized gradient search algorithm that can achieve global optimization by monotonically reducing the quantization step with respect to time when quantization is composed of integer or fixed-point fractional values applied to an optimization algorithm. According to the white noise hypothesis states, a quantization step is sufficiently small and the quantization is well defined, the round-off error caused by quantization can be regarded as a random variable with identically independent distribution. Thus, we rewrite the searching equation based on a gradient descent as a stochastic differential equation and obtain the monotonically decreasing rate of the quantization step, enabling the global optimization by stochastic analysis for deriving an objective function. Consequently, when the search equation is quantized by a monotonically decreasing quantization step, which suitably reduces the round-off error, we can derive the searching algorithm evolving from an optimization algorithm. Numerical simulations indicate that due to the property of quantization-based global optimization, the proposed algorithm shows better optimization performance on a search space to each iteration than the conventional algorithm with a higher success rate and fewer iterations.

• Structural Engineering and Mechanics
• /
• v.16 no.2
• /
• pp.219-240
• /
• 2003

One of the most versatile approaches for analyzing the dynamic behavior of structural systems is direct time integration of semi-discrete equations of motion. However responses computed by time integration are generally inexact and hence the corresponding errors would rather be studied in advance. In spite of the various error estimation formulations that exist in the literature, it is accepted practice to repeat the analyses with smaller time steps, followed by a comparison between the results. In this paper, after a review of this simple method and disregarding the round-off errors, a more efficient, reliable and yet simple method for estimating errors and enhancing the accuracy is proposed. The main objectives of this research are more realistic error estimation based on the concept of convergence, approximately controlling the reliability by comparing the actual rate of convergence with the integration method's order of accuracy, and enhancement of reliability by applying Richardson's extrapolation. Starting from the errors at specific time instants, the study is then generalized to cases in which the errors should be estimated and decreased at specific events e.g. peak responses. Numerical study illustrates the efficacy of the proposed method.

• Journal of applied mathematics & informatics
• /
• v.7 no.2
• /
• pp.409-438
• /
• 2000

In this paper we develop a new procedure to control stepsize for Runge- Kutta methods applied to both ordinary differential equations and semi-explicit index 1 differential-algebraic equation In contrast to the standard approach, the error control mechanism presented here is based on monitoring and controlling both the local and global errors of Runge- Kutta formulas. As a result, Runge-Kutta methods with the local-global stepsize control solve differential of differential-algebraic equations with any prescribe accuracy (up to round-off errors)

• Journal of KIISE:Information Networking
• /
• v.29 no.3
• /
• pp.277-285
• /
• 2002

This paper proposes and analyzes an error control scheme for the transmission control based real-time communication, such as FDDI, TDMA, and wireless LAN, which delivers the message according to the round robin fashion after the off-line bandwidth allocation. Taking into account the time constraint of each message, the proposed error control scheme makes the receiver transmit the error report via asynchronous traffic while the sender resend the requested message via overallocated access time which is inevitably introduced by the bandwidth allocation procedure for hard real -time guarantee. The error control procedure does not interfere other real -time message transmissions. In addition, as each frame contains the size of the message it belongs, the receiver can recognize the end of completion of message transmission. This enables earlier error report to the receiver so that the sender can cope with more network errors. The analysis results along with simulation performed via SMPL show that the proposed scheme is able to enhance the deadline meet ratio of messages by overcoming the network errors. Using the proposed error control scheme, the hard real -time network can be built at cost lower than, but performance comparable to the expensive dual link network.

• Computational Structural Engineering
• /
• v.3 no.1
• /
• pp.59-70
• /
• 1990

The p-version of the finite element method is a new approach to finite element analysis in which the partition of the domain is held fixed while the degree p of approximating piecewise polynomials is increased. In this paper, the focus is on computer implementation of a new hierarchic p-convergence shell model based on blend mapping functions. Its rigid-body modes, round-off error, and convergence characteristics are investigated.

• Korean Journal of Optics and Photonics
• /
• v.10 no.5
• /
• pp.405-412
• /
• 1999

We propose an effective algorithm, which can predict the detailed behavior of the intensity-modulated high speed optical signal after propagating through an optical fiber. The alogrithm is based on the SSF (Split Step Fourier) Method, however, the step size is automatically calibrated in each calculation step to reduce the number of calculations within given round-off error bound. Applying the algorithm to the 2.5 Gbps 100 km transmission and 10 Gbps 40 km transmission simulations, we achieved the calculation time reduction by maximum 1/120 and 1/56 of the calculation time by using the SSF fixed step algorithm previously known. The root-mean-square of the round-off error was kept within -30 dB compared to the signal level throughout the calculation.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.18 no.3
• /
• pp.624-634
• /
• 1994

A numerical method which combines equal-order velocity-pressure formulation originated from SIMPLE algorithm and streamline upwinding method has been developed. To verify the proposed numerical method, we considered the lid-driven cavity flow and backward facing step flow. The trend of convergence history is stable up to the error criterion beyond which the maximum value of error is oscillatory due4 to the round-off error. In the present study, all results were obtained with the single precision calculation up to the given error criterion and it was found to be sufficient for our purpose. The present results were then compared with existing experimental results using laser doppler velocimetry and numerical results using finite difference method and mixed interpolation finite element method. It has been shown that the present method gives accurate results with less memories and execution time than the coventional finite element method.