• International Journal of Control, Automation, and Systems
• /
• v.3 no.2
• /
• pp.202-210
• /
• 2005

This paper presents a new neural network control for nonlinear bridge systems with earthquake excitation. We design multi-layer neural network controllers with a single hidden layer. The selection of an optimal number of neurons in the hidden layer is an important design step for control performance. To select an optimal number of hidden neurons, we progressively add one hidden neuron and observe the change in a performance measure given by the weighted sum of the system error and the control force. The number of hidden neurons which minimizes the performance measure is selected for implementation. A neural network was trained for mitigating vibrations of bridge systems caused by El Centro earthquake. We applied the proposed control approach to a single-degree-of-freedom (SDOF) and a two-degree-of-freedom (TDOF) bridge system. We assessed the robustness of the control system using randomly generated earthquake excitations which were not used in training the neural network. Our results show that the neural network controller drastically mitigates the effect of the disturbance.

• The KIPS Transactions:PartC
• /
• v.11C no.2
• /
• pp.245-252
• /
• 2004

Stream Control Transmission Protocol(SCTP) is a new connection-oriented, reliable delivery transport protocol operating on top of an unreliable connectionless packet service such as IP. It inherits many of the functions developed for TCP, including flow control and packet loss recovery functions. In addition, it also supports transport layer multihoming and multistreaming In this paper, we study the impact of multi-homing on the performance of SCTP. We first compare performance of single-homed SCTP. multi-homed SCTP, TCP Reno and TCP SACK. We, then describe potential flaw in the current SCTP retransmission policy, when SCTP host is multihomed. Our Results show that SCTP performs better than TCP Reno and TCP SACK due to several changes from TCP in its congestion control mechanism. In particular. multi-homed SCTP shows the best result among the compared schemes. Through experimentation for multi-homed SCTP, we found that the current SCTP retransmission policy nay deteriorate the perfomance when the retransmission path it worse than the original path. Therefore, the condition of retransmission path is a very important factor In SCTP performance and a proper mechanism would be required to measure the condition of the retransmission path.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.3B
• /
• pp.130-137
• /
• 2005

Wireless communication environment is changing rapidly as we use new wireless communication technology such as WiBro to access high speed Internet. As a result, reliable data transmission using TCP is also expected to increase. Since TCP assumes that it is used in wired network, TCP suffers significant performance degradation over wireless network where packet losses are related to non-congestion loss. Especially RTO imposes a great performance degradation of TCP. In this paper, we analyze the loss recovery probabilities based on previous researches, and use simulation results of our algorithm to show that it prevents performance degradation by quickly detecting and recovery losses without RTO during fast recovery.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.04a
• /
• pp.484-486
• /
• 2002

Mobile IP는 이동단말의 macro-mobility를 지원하는 표준이다. 그런데, Mobile IP는 도메 인내에서의 빈번한 핸드오프상에서는 많은 단점이 있다. HA(Home Agent)로 빈번하게 시그널을 보내기 때문에 많은 제어 메시지 오버gp드가 있으며, 핸드오프중의 많은 패킷 손실이 있다. Micro-mobility프로토콜은 기존 mobile IP의 단점을 보완하기 위해 나온 프로토콜이다. 무선 네트웍상의 도메인내에서의 이동성을 지원하며, 손실과 메시지 오버 헤드를 줄여 핸드오프시 높은 성능향상을 주고, 효율적인 위치관리를 제공한다. TCP(Transmission Control Protocol)는 전통적인 네트웍에서 잘 동작하는 신뢰성 있는 연결지향성 전송프로토콜이다. 그러나, TCP는 Mobile IP를 이용한 무선환경하에서는 잦은 핸드오프 때문에 손실과 지연이 일어나서 많은 성능저하가 일어난다. 본 논문에서는 TCP Tahoe, Reno, NewReno, SACK를 micro-mobility 프로토콜상에서 사용하였을 때 어느 TCP버전과 micro-mobility프로토콜이 가장 효율적인지를 제시한다.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.408-413
• /
• 2005

A new methodology for discrete non-Gaussian probability density estimation is investigated in this paper based on a dynamic Bayesian network (DBN) and kernel functions. The estimator consists of a DBN in which the transition distribution is represented with kernel functions. The estimator parameters are determined through a recursive learning algorithm according to the maximum likelihood (ML) scheme. A discrete-type Poisson distribution is generated in a simulation experiment to evaluate the proposed method. In addition, an unknown probability density generated by nonlinear transformation of a Poisson random variable is simulated. Computer simulations numerically demonstrate that the method successfully estimates the unknown probability distribution function (PDF).

• Journal of Mechanical Science and Technology
• /
• v.20 no.5
• /
• pp.591-601
• /
• 2006

In this paper, we present a new control methodology for perturbed crane systems. Nonlinear crane systems are transformed to linear models by feedback linearization. An inverse dynamic equation is applied to compute the system PD control force. The PD control parameters are selected based on a nominal model and are therefore suboptimal for a perturbed system. To achieve the desired performance despite model perturbations, we construct a neural network auxiliary controller to compensate for modeling errors and disturbances. The overall control input is the sum of the nominal PD control and the neural auxiliary control. The neural network is iteratively trained with a perturbed system until acceptable performance is attained. We apply the proposed control scheme to 2- and 3-degree-of-freedom (D.O.F.) crane systems, with known bounds on the payload mass. The effectiveness of the control approach is numerically demonstrated through computer simulation experiments.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.12B
• /
• pp.1106-1114
• /
• 2003

Several analytic models describe transmission control protocol (TCP) performance such as steady-state throughput as an averaged ratio of number of transmissions to latency. For more detailed analysis of TCP latency, the latency during packet losses are recovered should be considered. In this paper, we derive the expected duration of loss recovery latency considering the number of packet losses recovered by retransmissions. Based on the numerical results verified by simulations, TCP using selective acknowledgement (SACK) option is more effective than TCP NewReno from the aspect of loss recovery latency.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.37 no.10
• /
• pp.18-25
• /
• 2000

In this paper, we analyze TCP congestion control algorithm over ATM-UBR network. TCP congestion control algorithm consists of slow start, congestion avoidance, fast recovery, fast retransmit. We analyze the ATM-UBR network service using the BSD 4.3 TCP Reno, Vanilla. However we found the fact that the characteristic of fast retransmit, recovery algorithm makes a serious degradation of Performance in multiple losses of packets. We propose new fast retransmit, recovery algorithm to improve the problem. The results of performance analysis improve the multiple losses of packets using a proposed fast retransmit, recovery algorithm.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04a
• /
• pp.637-639
• /
• 2004

GFR 서비스는 ATM 네트워크에서 인터넷 트래픽을 효과적으로 처리하기 위하여 ATM 포럼에서 정의되었다. 단순 FIFO 기반 GFR 구현 기법은 가상 연결 단위의 계수기나 분리된 버퍼 관리가 필요치 않아 실용적이지만, QoS가 만족스럽지 않다. 본 논문은 이 구현 기법의 QoS를 개선할 수 있는 방안을 찾기 위해, 이 구현 기법의 문제점을 명확히 하고자 컴퓨터 시뮬레이션을 수행하고 그 결과를 분석하였다. TCP 트래픽에 따른 성능 평가를 위해 Tahoe와 New-Reno 두 버전을 각각 적응하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.6B
• /
• pp.430-439
• /
• 2008

In this paper, we propose a sender-based TCP congestion control scheme for downward vertical handover (DVHO), in which mobile node moves from a cellular network to a wireless LAN. DVHO can give rise to severe performance problems in TCP throughput because it causes a drastic change of link characteristics. Particularly, TCP executes falsely congestion control by packet reordering, which is occurred from link delay difference between a cellular link and a wireless LAN link. Therefore, the congestion window is reduced. And unnecessary retransmissions wastes bandwidth. To solve these problems, we propose a method using estimated round-trip time in cellular link to process duplicated ACKs from reordering. Furthermore, the duplicated ACKs are used to the control congestion window size. Simulation result shows that the proposed scheme can solve problems. Moreover, the proposed scheme can have better performance than TCP New Reno and nodupack.