• The KIPS Transactions:PartA
• /
• v.9A no.4
• /
• pp.503-510
• /
• 2002

In this paper, a general purpose embedded computer system is designed and implemented. The most crucial points of the system are extensibility and flexibility. The hardware of the developed system is composed of three modules and the software provides hardware independent application program interfaces. Moreover, uC/OS-II, a well-known open realtime kernel. is extended and ported onto the system. The extended uC/OS-II supports that multiple processes can have the same priority and such processes are scheduled in a round-robin manner.

• The KIPS Transactions:PartA
• /
• v.14A no.7
• /
• pp.443-450
• /
• 2007

Performance scalability and storage scalability become important in a large scale cluster of wireless internet proxy cache servers. Performance scalability means that the whole performance of the cluster increases linearly according as servers are added. Storage scalability means that the total size of cache storage in the cluster is constant, regardless of the number of cache servers used, if the whole cache data are partitioned and each partition is stored in each server, respectively. The Round-Robin based load balancing method generally used in a large scale server cluster shows the performance scalability but no storage scalability because all the requested URL data need to be stored in each server. The hashing based load balancing method shows storage scalability because all the requested URL data are partitioned and each partition is stored in each server, respectively. but, it shows no performance scalability in case of uneven pattern of client requests or Hot-Spot. In this paper, we propose a novel dynamic hashing method with performance and storage scalability. In a time interval, the proposed scheme keeps to find some of requested URLs allocated to overloaded servers and dynamically reallocate them to other less-loaded servers. We performed experiments using 16 PCs and experimental results show that the proposed method has the performance and storage scalability as different from the existing hashing method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.9A
• /
• pp.702-709
• /
• 2003

In this paper, we propose an efficient and QoS-aware MAC scheduling algorithm for Bluetooth, which considers both throughput and delay performance of each Master-Slave pair in scheduling decisions, and thus, attempts to maximize overall performance. The proposed algorithm, MTDPP (Modified Throughput-Delay Priority Policy), makes up for the drawbacks of T-D PP (Throughput-Delay Priority Policy) proposed in [6] and improves the performance. Since Bluetooth employs a master-driven TDD based scheduling algorithm, which is basically operated with the Round Robin policy, many slots may be wasted by POLL or NULL packets when there is no data waiting for transmission in queues. To overcome this link wastage problem, several algorithms have been proposed. Among them, queue state-based priority policy and low power mode-based algorithm can perform with high throughput and reasonable fairness. However, their performances may depend on traffic characteristics, i.e., static or dynamic, and they require additional computational and signaling overheads. In order to tackle such problems, we propose a new scheduling algorithm. Performance of our proposed algorithm is evaluated with respect to throughput and delay. Simulation results show that overall performances can be improved by selecting suitable parameters of our algorithm.

• KIPS Transactions on Computer and Communication Systems
• /
• v.6 no.5
• /
• pp.219-230
• /
• 2017

General-Purpose Graphics Processing Units (GPGPUs) build massively parallel architecture and apply multithreading technology to explore parallelism. By using programming models like CUDA, and OpenCL, GPGPUs are becoming the best in exploiting plentiful thread-level parallelism caused by parallel applications. Unfortunately, modern GPGPU cannot efficiently utilize its available hardware resources for numerous general-purpose applications. One of the primary reasons is the inefficiency of existing warp/thread block schedulers in hiding long latency instructions, resulting in lost opportunity to improve the performance. This paper studies the effects of hardware thread scheduling policy on GPGPU performance. We propose a novel warp scheduling policy that can alleviate the drawbacks of the traditional round-robin policy. The proposed warp scheduler first classifies the warps of a thread block into two groups, warps with long latency and warps with short latency and then schedules the warps with long latency before the warps with short latency. Furthermore, to support the proposed warp scheduler, we also propose a supplemental technique that can dynamically reduce the number of streaming multiprocessors to which will be assigned thread blocks when encountering a high contention degree at the memory and interconnection network. Based on our experiments on a 15-streaming multiprocessor GPGPU platform, the proposed warp scheduling policy provides an average IPC improvement of 7.5% over the baseline round-robin warp scheduling policy. This paper also shows that the GPGPU performance can be improved by approximately 8.9% on average when the two proposed techniques are combined.

• The KIPS Transactions:PartC
• /
• v.10C no.2
• /
• pp.185-190
• /
• 2003

Bluetooth communication is based on piconet, which is composed by one master and maximum seven slaves. Several piconets can be interconnected via an inter-piconet Bluetooth unit called a bridge unit to form a Bluetooth scatternet. This bridge node can make its presence in each piconet by switching. This switching must be carefully scheduled so that slot wastage and, hence, packet delays are minimized. In this paper, we introduce an efficient inter-piconet scheduling scheme based on sniff mode. This scheme tries to minimize the wastage of slots by having the bridge unit sniff with its peering masters with time limits and communicate with its slaves in remaining slots. The sniff time limits are determined adaptively based on the amount of traffic in each piconet. Simulation results show this scheme outperforms round-robin scheme based on sniff intervals of equal lengths.

• Journal of Korea Multimedia Society
• /
• v.9 no.7
• /
• pp.827-834
• /
• 2006

In distributed virtual environment, system response time to users' interactions is an importance factor to determine the degree of contentment with the system. Generally, response time is affected by the efficiency of event message processing algorithm, and most of previous systems use FCFS algorithm, which processes message traffic sequentially based only on the event occurrence time. Since this method totally depends on the order of occurrence without considering the priorities of events, it has a problem that might drop the degree of system response time by causing to delay processing message traffic when a bottleneck phenomenon happens in the server side. To overcome this limitation of FCFS algorithm, this paper proposes a concurrent event scheduling algorithm, which is able to process event messages concurrently by assigning the priorities to the events. It is also able to satisfy the two goals of system together, consistency and responsiveness through the combination of occurrence time with priority concept of events.

• 대한공업교육학회지
• /
• v.32 no.2
• /
• pp.199-216
• /
• 2007

Stream Control Transmission Protocol(SCTP) is a transport layer protocol to support the data transmission. SCTP is similar to Transmission Control Protocol(TCP) in a variety of aspects. However, several features of SCTP including multi-homing and multi-streaming incur the performance difference from TCP. This paper highlights the data transfer during the initial slow start phase in SCTP congestion control composed of slow start phase and congestion avoidance phase. In order to compare the mean transfer time between SCTP and TCP, we experiment with different performance parameters including bandwidth, round trip time, and data length. By varying data length, we also measure the corresponding initial window size, which is one of factors affecting the mean transfer time. For the experiment, we have written server and client applications by C language using SCTP socket API and have measured the transfer time by ethereal program. We transferred data between client and server using round-robin method. Analysis of these experimental results from the testbed implementation shows that larger initial window size of SCTP than that of TCP brings the reduction in the mean transfer time of SCTP compared with TCP by 15 % on average during the initial slow start phase.

• Journal of KIISE:Databases
• /
• v.29 no.1
• /
• pp.15-26
• /
• 2002

Most work on multimedia storage systems has assumed that client will be serviced using a round-robin strategy. The server services the clients in rounds and each client is allocated a time slice within that round. Furthermore, most such algorithms are evaluated on the basis of a tightly coupled cost function. This is the basis of well-known algorithm such as FCFS, SCAN, SCAN-EDF, etc. In this paper, we describe a scheduling module called Request Unifier(RU) that takes as input, a set of client request, and a set of constraints on the desired performance such as client waiting time or maximum disk bandwidth, and a cost function. It produces as output a Unified Read Request(URR), telling the storage server which data items to read and when these data items to be delivered to the clients. Given a cost function, a URR is optimal if there is no other URR satisfying the constraints with a lower cost. We present three algorithms in this paper that can accomplish this kind of request merging and compare their performance through an experimental evaluation.

• Journal of KIISE:Computing Practices and Letters
• /
• v.8 no.2
• /
• pp.183-194
• /
• 2002

The clustering environment with heterogeneous workstations provides the cost effectiveness and usability for executing applications in parallel. Load balancing is considered a necessary feature for a cluster of heterogeneous workstations to minimize the turnaround time. Previously, static load balancing that assigns a predetermined weight for the processing capability of each workstation, or dynamic approaches which execute a benchmark program to get relative processing capability of each workstation were proposed. The execution of the benchmark program, which has nothing to do with the application being executed, consumes the computation time and the overall turnaround time is delayed. In this paper, we present efficient methods for task distribution and task migration, based on the relative load index. We designed and implemented a load balancing system for the clustering environment with heterogeneous workstations. Turnaround times of our methods and the round-robin approach, as well as the load balancing method using a benchmark program, were compared. The experimental results show that our methods outperform all the other methods that we compared.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.5C
• /
• pp.486-495
• /
• 2002

In this paper, we propose a delay reduction method for frame-based packet transmit scheduling algorithm. A high-speed network such as ATM network has to provide some performance guarantees such as bandwidth and delay bound. Framing strategy naturally guarantees bandwidth and enables simple rate-control while having the inherently bad delay characteristics. The proposed delay reduction method uses the same hierarchical frame structure as HRR (Hierarchical Round-Robin) but does not use the static priority scheme such as round-robin. Instead, we use a dynamic priority change scheme so that the delay unfairness between wide bandwidth connection and narrow bandwidth connection can be eliminated. That is, we use FIFO (First-In First-Out) concept to effectively reduce the occurrence of worst-case delay and to enhance delay distribution. We compare the performance for the proposed algorithm with that of HRR. The analytic and simulation results show that HFIFO inherits almost all merits of HRR with fairly better delay characteristics.