Recently, there have been studies on space and time priority queues, where space priorities are given to a class of packets that are sensitive to loss, and time priorities to another class of packets that are sensitive to delay. However, these studies have been restricted to such models with push-out space priorities. In this paper, we extend the studies to the space and time priority M/G/1 model with partial-buffer-sharing (PBS) space priorities, where the whole buffer is divided into two regions: one is shared by packets of all classes and the other is dedicated only for packets of the higher space-priority class. Since the PBS space-priority mechanism can be implemented more readily in communication systems than the push-out one, there have been a lot of contributions on PBS space-priority queues. However, there are no contributions on space and time priority queues with PBS space priorities. To analyze the proposed queueing model, we first study the probabilistic structure of the service time of a packet, which is more involved to analyze than the push-out alternative because it may be divided into three different regimes: a regime (S-period) from the beginning of the service until the shared buffer region becomes full, a second one (P-period) from the end of the S-period until the whole buffer becomes full, a third one (F-period) from the end of the P-period until the end of the service. Using the distributions of the S-, P-, F-periods, we then construct and analyze the embedded Markov chain and the corresponding semi-Markov process governing the system state, and also derive system performance measures such as expected sojourn times and loss probabilities of different priority classes of packets. In numerical examples, we finally explore the effect of the shared buffer size, which is a major system control parameter of PBS priority queues, and the distributions of the service times of packets of different classes on the system performance measures.

1. 서 론
2. 모 형
3. 서비스 시간의 확률적 구조
4. 대기행렬 모형의 EMC와 SMP 분석
5. 성능 지표
6. 수치 예제
7. 결 론
References

• Kilhwan Kim(Department of Management Engineering, Sangmyung University) | 김길환 (상명대학교 경영공학과) Corresponding author