• The KIPS Transactions:PartC
• /
• v.12C no.5 s.101
• /
• pp.679-686
• /
• 2005

Because EPON access network shares a medium and aggregates the traffic from EPON subscribers, scheduling media access control on EPON bandwidth allocation is very important. Furthermore DBA mechanism of EPON based on TDMA is out of specification and up to implementation. This paper deals with a DBA method to guarantee the QoS of the delay sensitive traffic on the base of best-effort service and delay priority queue management. The proposed method performs virtual scheduling algorithm for the integrated traffic. It uses the same MAC messages and tries to guarantee the QoS of higher priority traffic first with a simple DBA architecture. We evaluate the algorithm for traffic delay according to polling interval and traffic load of upstream and downstream. The results show that the proposed method can guarantee the QoS of the delay sensitive traffic with priority of the service classes.

• The KIPS Transactions:PartA
• /
• v.13A no.3 s.100
• /
• pp.261-266
• /
• 2006

Priority queues can be used in applications such as scheduling, sorting, and shortest path network problem. Fibonacci heap, pairing heap, and M-heap are priority queues based on pointers. This paper proposes a modified M-heap with an way representation, called MA-heap, that resolves the problem mentioned in [1]. The MA-heap takes O(1) amortized time and O(logn) time to insert an element and delete the max/min element, respectively. These time complexities are the same as those of the M-heap. In addition, it is much easier to implement an MA-heap than a heap proposed in [5] since it is based on the simple traditional heap.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.1
• /
• pp.237-245
• /
• 1997

In this paper, a flow control mechanism is proposed which is based on the priority control between communication path of a node. In this scheme, demanding length of a data queue for any pre-defined, then each node in that path is forced to maintains buffer size under the limit by controlling priority level of the path. The communication path which requires higher bandwidth sets its demanding queue length smaller. By providing relationship between the priority of a path and length of its queue, the high bandwidth requesting path has a better chance to get high bandwidth by defining the smaller demanding queue size. And also, by forcing a path which has high flow rate to maintain small queue size in the path of the communication, the scheme keep the transmission delay of the path small. The size of the demanding queue of a path is regularly adjusted to meet the applications requirement, and the load status of the network during the life time of the communication. The priority control based on the demanding queue size is also provided in the intermediate nodes as well as the end nodes. By that the flow control can provide a quicker result than end to-end flow control, it provides better performance advantage especially for the high speed network.

• The KIPS Transactions:PartC
• /
• v.15C no.5
• /
• pp.419-428
• /
• 2008

IEEE 802.15.4 reserves transmission time to support real-time transport by sending GTS request packets to the PAN coordinator in advance. This paper introduces GTS-FAT technique to reduce the reservation time by giving a higher sending priority to GTS request packets than data packets. Differently from the conventional scheme where these two kinds of packets share a single transmission queue, GTS-FAT scheme allocates two queues with two different contention window sizes like IEEE 802.11e. This paper also proposes an analytical GTS delay model by combining the two legacy models for 802.15.4 and 802.11e to accurately predict the GTS-FAT delay over a given network topology. Our analysis shows that GTS-FAT reduces GTS service delay by up to 50% at the expense of the data delay by only up to 6.1% when GTS request packets four times outnumber data packets.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.7
• /
• pp.1363-1370
• /
• 2009

Recently, the automobile industry is going through drastic environmental changes. The relative importance of information technology rapidly developed so far is getting heavier as it is grafted to electrical and electronic areas among all other automobile-related technologies. In this paper, an improved algorithm from the real-time scheduling algorithm of operation system which is loaded in embedded system will be presented. The number of wait-queue of priority was reduced from 16 to 4 in the parts where wide differences were shown in scheduling algorithm of the existing OSEK OS. While the FIFO algorithm was used in wait-queue, the EDF algorithm was applied to the proposed scheduling algorithm, which more improved the real-timeness. Also a simple experiment on the proposed scheduling algorithm was conducted.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.05b
• /
• pp.1047-1050
• /
• 2003

최근 HAVi의 기반이 되는 IEEE1394는 기존의 멀티미디어 데이터를 지원하는 장비의 인터페이스 이외에도 실시간성이 요구되는 백색가전 제어 시스템에 적용이 되고 있으며 그 응용분야가 점차 확대되고 있다. HAVi는 순차적인 방식으로 메시지를 전달하므로 시급한 이벤트가 발생된 경우 기존의 많은 메시지로 인해 그 처리가 지연되어 문제가 발생될 수 있다. 따라서 본 논문에서는 우선순위 기반의 메시지 전송 기능을 추가한 HAVi 시스템의 설계 및 구현에 대해 기술한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.10c
• /
• pp.130-132
• /
• 2001

멀티미디어 통신을 위한 계층형 멀티캐스트를 무선 환경으로 화장할 때에는 무선 네트워크의 낮은 대역폭에 의한 네트워크 혼잡 및 핸드오프에 의한 전송률 불안정 현상이 나타나게 된다. 이러한 문제를 효과적이고 능동적으로 해결하기 위해 본 논문에서는이동 지원국 (MSS) 을 이용한 능동적인 전송률 조절 기법과 우선 순위에 따른 패킷 누락 기법을 제안한다. MSS는 네트워크 계층의 전송 큐 사용량을 통해 현재의 네트워크 상태를 파악해, 네트워크 혼잡 시 이동 호스트(MH)의 가입 실험 (Join-experiment) 을 막거나 수신 계층의 제거를 종용할 수 있다. 그리고 MSS에서 직접 계층별 우선 순위를 부여함으로써 핸드오프에 의한 전송량 불안정성 및 대역폭 분배 불공평성을 해결하고자 한다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.7A
• /
• pp.1170-1181
• /
• 2001

본 논문은 다양한 실시간 스케줄링 알고리즘과 자원 접근 제어 정책을 통합적으로 설계/구현하되 필요에 따라 시스템을 선택적으로 재구성할 수 있게 하는 통합 실시간 스케줄링 모델을 제안한다. 제안 모델은 [3, 4]에서 제안된 기존 모델에 자원 관리자 및 대기 큐 관리자가 추가되었다. 사용자는 태스크 및 자원 속성을 기반 스케줄링 알고리즘에 상관없이 동일하게 지정할 수 있다. 반면 시스템 설계자는 우선순위 역전 문제를 해결하고 공유 자원에 대한 한정된 블록킹 시간을 보장하기 위한 다양한 자원 접근 제어 정책들을 하위 단계의 복잡한 커널 모듈을 수정하지 않고도 효율적으로 개발할 수 있다. Real-Time Linux [6]에 제안된 스케줄러 모델을 구현한 후, 이를 기반으로 다양한 스케줄링 알고리즘과 자원 접근 제어 정책들을 시험적으로 구현하여 보았다. 여러 성능 실험을 통해 제안 모델을 기반으로 다양한 알고리즘과 정책을 구현한다 해도 실행시의 오버헤드는 크지 않은 반면, 시스템 재구성과 자원 접근 제어 정책을 효과적으로 지원할 수 있다는 것을 확인할 수 있었다.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.45 no.2
• /
• pp.55-60
• /
• 2008

Packet-switched wavelength division multiplexing (WDM) ring networks have been extensively studied as solutions to the increasing amount of traffic in metropolitan area networks, which is widely expected to be a mix of unicast and multicast traffic. In this paper we study the fairness between unicasting and multicasting in slotted packet-switched WDM ring networks that employ a tunable transmitter and fixed tuned receiver at each node and a posteriori buffer selection. We find that single-step longest queue selection algorithm generally results in unfairness between unicasting and multicasting or a lilted relative priority for multicast vs. unicast traffic. We present the various virtual output queue structures and their performance.

• Journal of the Korea Society for Simulation
• /
• v.18 no.4
• /
• pp.29-37
• /
• 2009

This paper deals with the priority queues exploited for the management of future event list in discrete event simulation. Among several implementations of priority queues, MList which consists of 3 tiers has been known to reveal the good performance. To improve the performance of MList, Dynamic-Shift MList (DSMList) is proposed in this paper. Whenever the number of events in tier 3 exceeds a critical number, DSMList creates new calendar queue in tier 2, then moves events from Tier 3 to the calendar queue. Instead of one calendar queue, therefore, a number of calendar queues are dynamically created in tier 2. Throughout experiments for the performance evaluation of DSMList, it shows that at least 20% improvement is obtained compared with MList.