• KIISE Transactions on Computing Practices
• /
• v.21 no.9
• /
• pp.604-609
• /
• 2015

The use of Operating System(OS) virtualization is increasing as it provides many useful features such as efficient use of hardware(HW), easy system migration, and isolation between virtual spaces which prevents errors effecting each other. Recent development in HW has made it possible to use OS virtualization in embedded systems. However, implementing OS virtualization means that a multiple number of schedulers are layered in a system, rendering it difficult to analyze the schedulability of the system and errors are easily produced. Errors in safety critical embedded systems can cause serious damage to life and property; thus, the hierarchical schedulability must be verified. In this paper, we propose a framework which supports formal modeling and verification of hierarchical scheduling systems with UPPAAL.

• The KIPS Transactions:PartB
• /
• v.12B no.2 s.98
• /
• pp.173-180
• /
• 2005

In this paper, we propose an efficient real-time scheduling method of multimedia tasks for decoding frames of MPEG-2 video streams. In our task model, each frame is decoded by a separate multimedia task. The decoding task for each frame is assigned to the priority according to the precedence and importance of frames in a video stream. We use a priority-based scheduling policy in order to effectively allocate the CPU bandwidth to multimedia tasks for MPEG-2 decoding. We show how to dynamically control the fraction of the CPU bandwidth allocated to each multimedia task according to the priority. The primary purpose of our scheduling method is to enhance the real-time performance of the multimedia system by minimizing the number of decoding tasks that have missed their deadlines while reducing the decoding times of these multimedia tasks. The performance of this scheduling method is compared with that of similar mechanisms through simulation experiments.

• The KIPS Transactions:PartC
• /
• v.18C no.6
• /
• pp.433-440
• /
• 2011

In this paper, we propose a new scheduling method that can guarantee reliable jitter by minimizing the queue length variation in the streaming service provisioning such as IPTV and VoD. The amount of traffic to be delivered within a certain time is very fluid because MPEG-4 and H.264 encoders use VBR(Variable Bit Rate) for delivering video streaming traffic. This VBR characteristic increases the end-to-end propagation delay variation when existing scheduling methods are used for delivering video frames. Therefore, we propose the new scheduling method that can minimize change rate of queue length by adaptively controling service rate taking into account the size of bulk incoming packets and arrival rate for bulk streaming traffic. Video frames can be more reliably transmitted through the minimization of the queue length variation using the proposed method. We use the queueing model and also carry out OPNET simulation to validate the proposed method.

• Journal of the Korea Society of Computer and Information
• /
• v.12 no.6
• /
• pp.1-9
• /
• 2007

We propose a novel framework for distributed resource management. The framework has the following novel features. First, the resource management system is distributed using resource content information that is characterized by system properties. We argue that a distributed system based on resource content is sufficient to satisfy specific scheduling requests for Quality of Service(QoS) considering workload balance across a grid. Second, the distributed system constructs a hierarchical peer-to-peer network. This peered network provides an efficient message routing mechanism. The simulation results demonstrate that the proposed framework is proficient to satisfy QoS in distributed environment.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.04a
• /
• pp.139-142
• /
• 2015

데이터 센터에서는 물리적인 규모 증가와는 달리 별도의 처리 없이는 분산처리 프레임워크가 동일한 클러스터 내에서 복수로 동작할 수 없어 전체 환경을 정적으로 분할하여 이들을 배치하는 것이 일반적이다. 그러나 최근 연구에서는 복수의 프레임워크를 한 클러스터 내에서 동작시킴으로써 클러스터의 활용률을 높이는 방향으로 이루어지고 있다. Mesos는 복수의 분산처리 프레임워크를 한 클러스터에서 동작시키기 위한 시스템 중 하나로 각 프레임워크 스케줄러의 스케줄링을 지원하는 단일 Allocation Module을 가진다. Allocation Module은 모든 Slave와 프레임워크 스케줄러들의 요청을 처리하는데, 시스템 규모가 커질수록 Allocation Module으로 집중되는 부하가 증가하여 이에 따른 할당 속도 저하로 정상적인 동작이 불가능해진다. 이 문제를 해결하기 위해 본 논문에서는 Mesos 시스템의 Allocation Module 병렬화를 제안한다. 제안 방식을 통해 Allocation Module의 부하를 분산함과 동시에 Head-of-line Blocking으로 인한 스케줄링 지연 문제를 해결할 수 있을 것이다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39C no.7
• /
• pp.538-545
• /
• 2014

This paper proposes a coordinated multi-point (CoMP) based dynamic transmission scheme in a downlink multi-ship network, where a central ship selects a ship in order to maximize the utility function. The proposed scheduling scheme dynamically decides to the usage of the coordinated multi transmissions and selects a user to be served for every frame, in order to the utility function on the basis of the throughput and fairness. In particular, the proposed utilify function based scheduling scheme aims to increase the quality of service of ships at the edge of cells. Under the proportional fair scheduling, the simulation results show that the proposed utility function-based scheduling improves the throughput of the ships at the cell edge with the little sacrifice of the system capacity.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04a
• /
• pp.151-153
• /
• 2003

이 논문은 프레임 기반의 비디오 응용에 대해서 화질 (video quality) 제약을 고려한 응용 수준(application-level)에서의 에너지 최적화 기법을 제안한다. 화질과 에너지 소모 사이에는 상관관계 (trade-off)가 있음을 이용하여. 본 논문은 H.263 인코더의 화질을 실시간으로 모니터링해서 프레임 속도(frame rate)를 자동으로 조절하는 알고리즘을 제안한다. 기존 동적 전압 스케줄링 (DVS: dynamic voltage scheduling) 기법은 유휴 시간 (slack time)을 주어진 것이라고 가정하는 반면, 제안하는 기법은 유휴 시간 자체를 증가시켜서 DVS 기법의 효과를 향상시킨다. 제안하는 기법이 주어진 화질 제약을 만족하며 상당한 에너지 소모를 감소함을 실험을 통해서 알 수 있다.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.3
• /
• pp.113-119
• /
• 2009

In this paper, we proposed an efficient TDMA scheduling algorithm for a clustered Wireless Sensor Network. Since some previous algorithms used unnecessary idle period and schedule period in each frame. It became an overhead that might consume unexpected energy and delay data transmission. To solve this problem, a dynamic scheduling algorithm according to the number of member nodes and node traffic load within a cluster was suggested. Our proposed DS-MAC(Dynamic Scheduling MAC) could save energy and reduce transmission delay Then DS-MAC was analyzed mathematically to compare with the previous algorithms.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.3
• /
• pp.63-67
• /
• 2024

Recently, artificial intelligence (AI) workloads encompassing various industries such as smart logistics, FinTech, and entertainment are being executed on the cloud. In this paper, we address the scheduling issues of various AI workloads on a multi-tenant cloud system composed of heterogeneous GPU clusters. Traditional scheduling decreases GPU utilization in such environments, degrading system performance significantly. To resolve these issues, we present a new scheduling approach utilizing genetic algorithm-based optimization techniques, implemented within a process-based event simulation framework. Trace driven simulations with diverse AI workload traces collected from Alibaba's MLaaS cluster demonstrate that the proposed scheduling improves GPU utilization compared to conventional scheduling significantly.

• Journal of Advanced Navigation Technology
• /
• v.2 no.1
• /
• pp.61-67
• /
• 1998

In this paper, we have described the Ramp Activity Coordination Expert System (RACES) which can solve aircraft parking problems. RACES includes a knowledge-based scheduling problem which assigns every daily arriving and departing flight to the gates and remote spots with the domain specific knowledge and heuristics acquired from human experts. RACES processes complex scheduling problem such as dynamic inter-relations among the characteristics of remote spots/gates and aircraft with various other constraints, for example, custome and ground handling factors at an airport. By user-driven modeling for end users and knowledge-driven near optimal scheduling acquired from human experts, RACES can produce parking schedules of aircraft in about 20 seconds for about 400 daily flights, whereas it normally takes about 4 to 5 hours by human experts. Scheduling results in the form of Gantt charts produced by the RACES are also accepted by the domain experts. RACES is also designed to deal with the partial adjustment of the schedule when unexpected events occur. After daily scheduling is completed, the messages for aircraft changes and delay messages are reflected and updated into the schedule according to the knowledge of the domain experts. By analyzing the knowledge model of the domain expert, the reactive scheduling steps are effectively represented as rules and the scenarios of the Graphic User Interfaces (GUI) are designed. Since the modification of the aircraft dispositions such as aircraft changes and cancellations of flights are reflected to the current schedule, the modification should be notified to RACES from the mainframe for the reactive scheduling. The adjustments of the schedule are made semi-automatically by RACES since there are many irregularities in dealing with the partial rescheduling.