• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 2004.07a
• /
• pp.74-74
• /
• 2004

• Proceedings of the KIEE Conference
• /
• 2005.10b
• /
• pp.325-327
• /
• 2005

In this paper, we propose the design of active queue management (AQM) control system using the self-recurrent wavelet neural network (SRWNN). By regulating the queue length close to reference value, AQM can control the congestions in TCP network. The SRWNN is designed to perform as a feedback controller for TCP dynamics. The parameters of network are tunes to minimize the difference between the queue length of TCP dynamic model and the output of SRWNN using gradient-descent method. We evaluate the performances of the proposed AQM approach through computer simulations.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.22 no.2
• /
• pp.45-65
• /
• 1997

We build a queueing model for buffered leaky bucket system. First, we set up system equations and them calculate the steady-state probabilities at an arbitrary time epoch by recursive method. We derive the mean waiting time and the mean number of cells in the input buffer, and evaluate the performance of the buffered leaky bucket system to find the optimal queue capacity and token generation rate that meet the quality of service(QoS).

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.9B
• /
• pp.783-792
• /
• 2003

We consider a finite-buffer discrete-time queueing system with fixed-size bulk-service discipline: Geo/ $G^{B}$1/K＋B. The main purpose of this paper is to present a performance analysis of this system that has a wide range of applications in Asynchronous Transfer Mode (ATM) and other related telecommunication systems. For this purpose, we first derive the departure-epoch probabilities based on the embedded Markov chain method. Next, based on simple rate in and rate out argument, we present stable relationships for the steady-state probabilities of the queue length at different epochs: departure, random, and arrival. Finally, based on these relationships, we present various useful performance measures of interest such as the moments of number of packets in the system at three different epochs and the loss probability. The numerical results are presented for a deterministic service-time distribution - a case that has gained importance in recent years.s.

• Journal of the Korea Society for Simulation
• /
• v.25 no.3
• /
• pp.117-123
• /
• 2016

This paper presents inference methods for inner operations of a multi-server queue when historical data are limited or system observation is restricted. In a queueing system analysis, autocorrelated arrival and service processes increase the complexity of modeling. Accordingly, numerous analysis methods have been developed. In this paper, we introduce an inference method for specific situations when external observations exhibit autocorrelated structure and and observations of internal operations are difficult. We release an assumption of the previous method and provide lemma and theorem to guarantee the correctness of our proposed inference method. Using only external observations, our proposed method deduces the internal operation of a multi-server queue via non-parametric approach even when the service times are autocorrelated. The main internal inference measures are waiting times and service times of respective customers. We provide some numerical results to verify that our method performs as intended.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.3
• /
• pp.66-75
• /
• 2017

Priority disciplines are an important scheme for service systems to differentiate their services for different classes of customers. (N, n)-preemptive priority disciplines enable system engineers to fine-tune the performances of different classes of customers arriving to the system. Due to this virtue of controllability, (N, n)-preemptive priority queueing models can be applied to various types of systems in which the service performances of different classes of customers need to be adjusted for a complex objective. In this paper, we extend the existing (N, n)-preemptive resume and (N, n)-preemptive repeat-identical priority queueing models to the (N, n)-preemptive repeat-different priority queueing model. We derive the queue-length distributions in the M/G/1 queueing model with two classes of customers, under the (N, n)-preemptive repeat-different priority discipline. In order to derive the queue-length distributions, we employ an analysis of the effective service time of a low-priority customer, a delay cycle analysis, and a joint transformation method. We then derive the first and second moments of the queue lengths of high- and low-priority customers. We also present a numerical example for the first and second moments of the queue length of high- and low-priority customers. Through doing this, we show that, under the (N, n)-preemptive repeat-different priority discipline, the first and second moments of customers with high priority are bounded by some upper bounds, regardless of the service characteristics of customers with low priority. This property may help system engineers design such service systems that guarantee the mean and variance of delay for primary users under a certain bounds, when preempted services have to be restarted with another service time resampled from the same service time distribution.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.767-771
• /
• 2004

In this paper the performance of some control theory-based Active Queue Management (AQM) methods are improved and investigated through simulation. The simulation scenario is considered to be closely realistic, in which the traffic is mainly due to http sessions. The results show the best performance of PID controller with respect to the others in good queue regulation as well as high link utilization and low delay metrics.

• Journal of the Korea Society for Simulation
• /
• v.24 no.3
• /
• pp.27-35
• /
• 2015

We analyze an M/G/1/K queueing system with queue-length dependent service and arrival rates. There are a single server and a buffer with finite capacity K including a customer in service. The customers are served by a first-come-first-service basis. We put two thresholds $L_1$ and $L_2$($${\geq_-}L_1$$ ) on the buffer. If the queue length at the service initiation epoch is less than the threshold $L_1$, the service time of customers follows $S_1$ with a mean of ${\mu}_1$ and the arrival of customers follows a Poisson process with a rate of ${\lambda}_1$. When the queue length at the service initiation epoch is equal to or greater than $L_1$ and less than $L_2$, the service time is changed to $S_2$ with a mean of $${\mu}_2{\geq_-}{\mu}_1$$. The arrival rate is still ${\lambda}_1$. Finally, if the queue length at the service initiation epoch is greater than $L_2$, the arrival rate of customers are also changed to a value of $${\lambda}_2({\leq_-}{\lambda}_1)$$ and the mean of the service times is ${\mu}_2$. By using the embedded Markov chain method, we derive queue length distribution at departure epochs. We also obtain the queue length distribution at an arbitrary time by the supplementary variable method. Finally, performance measures such as loss probability and mean waiting time are presented.

• Proceedings of the KIEE Conference
• /
• 2005.07d
• /
• pp.2720-2722
• /
• 2005

본 논문에서는 웨이블릿 신경 회로망에 기반을 둔 능동 큐 관리(Active Queue Management) 제어 시스템을 설계하는 것을 제안한다. 제안한 제어 시스템에서 웨이블릿 신경 회로망은 능동 큐 관리를 위한 제어기로 사용한다 TCP 동적 모델의 실제 출력, 큐의 길이와 웨이블릿 신경 회로망을 이용한 출력의 오차가 최소화가 되도록 웨이블릿 신경 회로망의 파라미터 값들을 변화시키며 각각의 파라미터 값들은 경사 하강법을 통해 학습시킨다. 마지막으로 제안한 방법은 모의실험을 통해 패킷 손실률과 큐의 길이의 관점에서 제안한 방법의 향상성을 보이고자 한다.

• ETRI Journal
• /
• v.31 no.1
• /
• pp.86-88
• /
• 2009

In this letter, we consider an m-node tandem queue (queues in series) with a Poisson arrival process and either deterministic or non-overlapping service times. With the assumption that each node has a finite buffer except for the first node, we show the non-increasing convex property of stationary waiting time with respect to the finite buffer capacities. We apply it to an optimization problem which determines the smallest buffer capacities subject to probabilistic constraints on stationary waiting times.