Atsushi Inoie

Stationary distribution of a multi-server vacation queue with constant impatient times

Abstract This paper analyzes a multi-server queueing system with multiple vacations in which the vacation duration of each server is exponentially distributed. When all the servers are on vacation, customers are impatient if their waiting times exceed a constant value. To derive the stationary distribution of the system, we employ the matrix analytic method. Specifically, our queueing model is represented as an M / G / 1 -type Markov chain, and the stationary distribution is simply obtained by using its transition structures. Furthermore, we explicitly derive the tail decay rate of the stationary distribution, and demonstrate that the decay rate is not always equal to that for the queueing model without vacations.

Simulation-Based Optimization Approach for an Extended-Kanban Control System

An approach to optimize the control parameters for an extended-kanban control system (EKCS) is studied. The EKCS is a Just-in-Time production system that involves the advantages of classical kanban and base-stock control systems. Although EKCS can be modeled as a discrete-time Markov chain, the state space is gigantic even for a single stage production system, making it impossible to numerically compute the Markov chain. Thus, we propose a simulation-based algorithm for parameter optimization.

Hybrid ABC algorithm for the capacitated vehicle routing problem

An effective meta-heuristic algorithm for a capacitated vehicle routing problem (CVRP) is studied. In this paper, we proposed the artificial bee colony (ABC) algorithm combined with simulated annealing (SA), and applied the algorithm to a simple CVRP model. We show the advantage of the proposed algorithm through some numerical experiments.

Tail asymptotics for waiting time distribution of an M/M/s queue with general impatient time

In this paper, we consider an M/M/s queue where customers may abandon waiting for service and renege the system without receiving their services. We assume that impatient time on waiting for each customer is independent and identically distributed non-negative random variable with a general distribution where the probability distribution is light-tailed and unbounded. The main objective of this paper is to provide an approximation for the waiting time distribution in an analytically tractable form. To this end, we obtain the tail asymptotics of the waiting time distributions of served and impatient customers. By using the tail asymptotics, we show that the fairly good approximations of the waiting time distributions can be obtained with low numerical complexity.

The Effect of Transmission Error on Audio Quality for Simple FEC Schemes

The audio quality provided by some simple FEC schemes is studied. We deal with an M/M/1/K queueing model and assume that some of packets are lost due to noise or interference in the transmission as well as buffer overflow in the router. We consider the following two schemes: One is the Fixed/Variable Scheme that fixes the volume of original data and varies the volume of redundant data in a packet, and the other is Variable/Fixed Scheme that varies the volume of original data and fixed the volume of the whole data. We showed the cases where the Fixed/Variable Scheme improves the audio quality of the system.

InfoShare : Information sharing system with a mechanism of selectively cooperative retrieval on hybrid P2P networks

We propose an information sharing system (called InfoShare) which is a set of multiple hybrid peer-to-peer (P2P) networks logically connected to a cooperative server. The design goal of the system is to form a multipurpose distributed database to share various kinds of information distributed on the campus and to access to the Internet website. Each of peers in the campus submits metadata of its own files to a meta-server (called information server) that forms a local database. As each local database may classify those information into different categories, the cooperative server plays an important role to succeed in reconciling paradox among them. When a certain peer retrieves information from the system, the information servers that have metadata included in the query cooperatively search target files. Consequently the peer gets the target file directly from another peer or a website. InfoShare also has a cache mechanism for well-referred files as a countermeasure taken to avoid peerpsilas leaving. Our experimental results show that InfoShare has an excellent retrieval response time and can be effectively adaptable to many applications of campus information sharing.