• Journal of applied mathematics & informatics
• /
• v.22 no.1_2
• /
• pp.273-283
• /
• 2006

We obtain an asymptotic behavior of the loss probability for the GI/PH/1/K queue as K tends to infinity when the traffic intensity p is strictly less than one. It is shown that the loss probability tends to 0 at a geometric rate and that the decay rate is related to the matrix generating function describing the service completions during an interarrival time.

• Journal of applied mathematics & informatics
• /
• v.27 no.5_6
• /
• pp.1459-1463
• /
• 2009

Recently, Zhao et.al [European Journal of Operational Research 169 (2006) 189-201] discussed a random fuzzy renewal process based on random fuzzy theory. They considered the rate of the random fuzzy renewal process and presented a random fuzzy elementary renewal theorem. They also established Blackwell's theorem in random fuzzy sense. But all these results are invalid. We give a counter example in this note.

• Journal of applied mathematics & informatics
• /
• v.28 no.1_2
• /
• pp.153-162
• /
• 2010

The finite buffer GI/G/1/K system is set up by using an unconventional arrangement of the state space, in which the remaining interarrival time or service time is chosen as the level. The stationary distributions of resulting Markov chain can be explicitly determined, and the chain is positive recurrent without any restriction. This is an advantage of this method, compared with that using the elapsed time approach [2].

• Journal of the military operations research society of Korea
• /
• v.12 no.2
• /
• pp.68-81
• /
• 1986

This thesis studies the estimation from interarrival and service time data of the mean number of customers in service at time t for an $M/G/{\infty}$ queue. The assumption is that the parametric form of the service time distribution is unknown and the empirical distribution of twe service time is used in the estimate the mean number of customers in service. In the case in which the customer arrival rate is known the distribution of the estimate is derived and an approximate normal confidence interval procedure is suggested. The use of the nonparametric methods, which are the jackknife and the bootstrap, to estimate variability and construct confidence intervals for the estimate is also studied both analytically and by simulation.

• Journal of Korean Institute of Industrial Engineers
• /
• v.20 no.4
• /
• pp.37-50
• /
• 1994

The efficiency of production system is mainly considered in the viewpoints of reducing the average flow time of products and increasing the throughput rate. The performance in these viewpoints is very depending on job dispatching of each machine in real time operation, in the case jobs are released dynamically. In this research, a heuristic dynamic dispatching method is suggested for a flow shop case where new jobs with random process times are released by an interarrival time distribution and the number of waiting jobs between each pair of machines are limited. The proposed method has been compared with some priority rule-based dispatching methods by simulation and found to be superior to them.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.12C
• /
• pp.199-207
• /
• 2001

We propose an adaptive shaping controller equipped with the technologies of shaping and buffering VoIP packets arriving at the receiving end by the CAM-type controller. In order to conceal interarrival packet delay variation, the conventional jitter buffers force them to be too large, thereby causing the audio quality to suffer excessive delay. However, by using our proposed method, the delay caused by shaping operation dynamically increases or decreases on the level of jitter that exists with in the IP network. This makes the delay accommodates adaptively the network jitter condition. The less jitter network has the fewer delay the shaping controller requires for jitter elimination. And the CAM-type method generally makes the shaping operation faster and leads to processing packets in as little time as can. We analyse the packet loss and delay performance dependency on the average talk ratio and the number of jitter buffer entries in shaping controller. Surprising, we show that the average delay using our shaping controller is about 70msec. This performance is much better than with the delay equalization method which forces the receiving end to delay about 60msec.

• Proceedings of the Korea Society for Simulation Conference
• /
• 1998.10a
• /
• pp.17-20
• /
• 1998

The purpose of this paper is to develop a port simulation program for Inchon Port. The arrival and departure data from Inchon Port are analyzed, several probability density functions for the interval time and service time are estimated, and the berthing rules are constructed based on experts experiences. The port simulation program is written in visual basic considering the interarrival p.d.f., service rate p.d.f., berthing rules, and dual-dock system, and is used to analyze the effects of the changes in ship arrival rate and service rate upon the demurrage. It is expected that we can use the simulation results in order to prepare proper service level and to evaluate the appropriate investment strategy to be planned.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.21 no.9
• /
• pp.2373-2384
• /
• 1996

Connection caching technique is widely used to reduce the overhead incurred by frequency connection establishment in distributed computing environment using connection oriented protocol. We present an efficient connection caching scheme where we divide the system into several host groups and connection between the hosts in the same group is kept prior to others. In other words it makes the connection kept longer by sharing the connection use information between hosts in the same group. Every host group consists of hosts which have heavy intercommunication. And we present performance evaluation of the effect of host grouping upon connection caching and performance comparison of various grouping in several aspects includingmeanservice request interarrival time, mean service time, group size, grouping and type of group. Simulation results show that host grouping is effective in evrey performance criterion and proper grouping of hosts enhances the performance. We also present analysis results for Markov process model of our scheme which are consistent with the simulation results.

• Journal of Communications and Networks
• /
• v.7 no.1
• /
• pp.1-12
• /
• 2005

Ethernet is being deployed in metropolitan area networks (MANs) as a lower-cost alternative to SONET-based infrastructures. MANs are usually required to support common communication services, such as voice and frame relay, based on legacy synchronous TDM technology in addition to asynchronous packet data transport. This paper addresses the clock synchronization problem that arises when transporting synchronous services over an asynchronous packet infrastructure, such as Ethernet. A novel algorithm for clock synchronization is presented combining time-stamp methods used in the network time protocol (NTP) with signal processing techniques applied to measured packet interarrival times. The algorithm achieves the frequency accuracy, stability, low drift, holdover performance, and rapid convergence required for viable emulation of TDM circuit services over Ethernet.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.1B
• /
• pp.17-24
• /
• 2008

The effect of the variable packet size on the LRD characteristic of the MMPP traffic model is investigated. When we generate packet traffic for the performance evaluation of IP packet network, MMPP model can be used to generate packet interarrival time. And a random length of packet size from a certain distribution can be assigned to each packet. However, there is a possibility that the variable packet size might change the LRD characteristic of the original MMPP model. In this study, we investigate this possibility. For this purpose the 'refined traffic' is defined, where packet arrival time is generated according to the MMPP model and a random packet length from a specific distribution is assigned to each generated packet. Hurst parameter of the refined traffic is estimated and compared with the original Hurst parameter, which is the input parameter of the MMPP model. We also investigate the effect of the packet size distribution on the queueing performance of the MMPP traffic model and the relationship between the Hurst parameter and queueing performance.