• Communications of the Korean Mathematical Society
• /
• v.14 no.3
• /
• pp.611-625
• /
• 1999

We consider ΜΑΡ/G/ 1 finite capacity queue with mul-tiple thresholds on buffer. The arrival of customers follows a Markov-ian arrival process(MAP). The service time of a customer depends on the queue length at service initiation of the customer. By using the embeded Markov chain method and the supplementary variable method, we obtain the queue length distribution ar departure epochs and at arbitrary epochs. This gives the loss probability and the mean waiting time by Little's law. We also give a simple numerical examples to apply the overload control in packetized networks.

• Journal of applied mathematics & informatics
• /
• v.8 no.1
• /
• pp.285-295
• /
• 2001

We consider a queueing system under overload control to support bursty traffic. The queueing system under overload control is modelled by MMBP/D1/K queue with two thresholds on buffer. Arrival of customer is assumed to be a Markov-modulated Bernoulli process (MMBP) by considering burstiness of traffic. Analysis is done in discrete-time case. Using the generating function method, we obtain the stationary queue length distribution. Finally, the loss probability and the waiting time distribution of a customer are given.

• Journal of Korea Society of Industrial Information Systems
• /
• v.23 no.1
• /
• pp.53-63
• /
• 2018

In this paper, we analyze the waiting time of a queueing system with D-BMAP (discrete-time batch Markovian arrival process) and D-policy. Customer group or packets arrives at the system according to discrete-time Markovian arrival process, and an idle single server becomes busy when the total service time of waiting customer group exceeds the predetermined workload threshold D. Once the server starts busy period, the server provides service until there is no customer in the system. The steady-state waiting time distribution is derived in the form of a generating function. Mean waiting time is derived as a performance measure. Simulation is also performed for the purpose of verification and validation. Two simple numerical examples are shown.

• Industrial Engineering and Management Systems
• /
• v.15 no.2
• /
• pp.123-130
• /
• 2016

We introduce a queueing system with general arrival stream and exponential service time under the N-policy, where customers may renege during idle period and arrival rates may vary according to the server's status. Probability distributions of the lengths of idle period and busy period are derived using absorbing Markov chain approach and a method to obtain the optimal control policy that minimizes long-run expected operating cost per unit time is provided. Numerical analysis is done to illustrate and characterize the method.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.38 no.2
• /
• pp.91-100
• /
• 2015

In this study, we propose an efficient two-phase heuristic policy, called an acceptance tolerance control policy, for Infrastructure as a Service (IaaS) cloud services that considers both the service provider and customer in terms of profit and satisfaction, respectively. Each time an IaaS cloud service is requested, this policy determines whether the service is accepted or rejected by calculating the potential for realizing the two performance objectives. Moreover, it uses acceptance tolerance to identify the possibility for error with the chosen decision while compensating for both future fluctuations in customer demand and error possibilities based on past decisions. We conducted a numerical experiment to verify the performance of the proposed policy using several actual IaaS cloud service specifications and comparing it with other heuristics.

• Journal of Korea Society of Industrial Information Systems
• /
• v.21 no.6
• /
• pp.1-12
• /
• 2016

In this paper, we consider a general discrete-time queueing system with D-BMAP(discrete-time batch Markovian arrival process) and D-policy. An idle single server becomes busy when the total service times of waiting customer group exceeds the predetermined workload threshold D. Once the server starts busy period, the server provides service until there is no customer in the system. The steady-state workload distribution is derived in the form of generating function. Mean workload is derived as a performance measure. Simulation is also performed for the purpose of verification and a simple numerical example is shown.

• Communications for Statistical Applications and Methods
• /
• v.25 no.4
• /
• pp.443-451
• /
• 2018

$P^M_{\lambda}$-service policy is a workload dependent hysteretic policy. The policy has two service states comprised of the ordinary stage and the fast stage. An ordinary service stage is initiated by the arrival of a customer in an idle state. When the workload of the server surpasses threshold ${\lambda}$, the ordinary service stage changes to the fast service state, and it continues until the system is empty. These service stages alternate in this manner. When the cost of changing service stages is high, the hysteretic policy is more efficient than the threshold policy, where a service stage changes immediately into the other service stage at either case of the workload's surpassing or crossing down a threshold. $P^M_{\lambda}$-service policy is a modification of $P^M_{\lambda}$-policy proposed to control finite dams, and also an extension of the well-known D-policy. The distributions of the stationary workload of $P^M_{\lambda}$-service policy and its variants are studied well. However, there is no known result on the sojourn time distribution. We prove that there is a relation between the sojourn time of a customer and the first up-crossing time of the workload process over the threshold ${\lambda}$ after the arrival of the customer. Using the relation and the duality of M/G/1 and G/M/1 queues, we obtain conditional sojourn time distributions in M/G/1 and G/M/1 queues under the policy.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.24 no.4
• /
• pp.111-122
• /
• 1999

In this paper, we consider a queueing network model. where the population constraint within each subnetwork is controlled by a semaphore queue. The total number of customers that may be present in the subnetwork can not exceed a given value. Each node has a constant service time and the arrival process to the queueing network is an arbitrary distribution. A major characteristics of this model is that the lower layer flow is halted by the state of higher layer. We present some properties that the inter-change of nodes does not make any difference to customer's waiting time in the queueing network under a certain condition. The queueing network can be transformed into a simplified queueing network. A dramatic simplification of the queueing network is shown. It is interesting to see how the simplification developed for sliding window flow control, can be applied to multi-layered queueing network.

• Journal of Korea Society of Industrial Information Systems
• /
• v.23 no.2
• /
• pp.29-39
• /
• 2018

In this paper, we discuss a discrete-time queueing system with dyadic server control policy that combines workload control and multiple vacations. Customers arrive at the system with Bernoulli arrival process. If there is no customer to serve in the system, an idle single server spends a vacation of discrete random variable V and returns. The server repeats the vacation until the total service time of waiting customers exceeds the predetermined workload threshold D. In this paper, we derived the steady-state workload distribution of a discrete-time queueing system which is operating under a more realistic and flexible server control policy. Mean workload is also derived as a performance measure. The results are basis for the analysis of system performance measures such as queue lengths, waiting time, and sojourn time.

• Journal of Korea Society of Industrial Information Systems
• /
• v.23 no.5
• /
• pp.19-32
• /
• 2018

This study focuses on an M/M/1 queue with an attached continuous-type inventory. The customers arrive into the system according to the Poisson process, and are served in their arrival order; i.e., first-come-first-served. The service times are assumed to be independent and identically distributed exponential random variable. At a service completion epoch, the customer consumes a random amount of inventory. The inventory is controlled by the traditional (s, S)-inventory policy with a generally distributed lead time. A customer that arrives during a stock-out period assumed to be lost. For the number of customers and the inventory size, we derive a product-form stationary joint probability distribution and provide some numerical examples. Besides, an operational strategy for the inventory that minimizes the long-term cost will also be discussed.