• IEMEK Journal of Embedded Systems and Applications
• /
• v.12 no.4
• /
• pp.193-204
• /
• 2017

Various 3D-stacked DRAMs have been proposed to overcome the memory wall problem. Hybrid Memory Cube (HMC) is a true 3D-stacked DRAM with stacked DRAM layers on top of a logic layer. The logic die is mainly used to implement a memory controller for HMC, and it is connected through a high speed serial link called SerDes with a host that is either a processor or another HMC. In HMC, the serial link is crucial for both performance and power consumption. Therefore, it is important that the link is configured properly so that the required performance should be satisfied while the power consumption is minimized. In this paper, we propose a HMC system model included the high speed serial link to estimate performance accurately. Since the link modeling strictly follows the link flow control mechanism defined in the HMC spec, the actual HMC performance can be estimated accurately with respect to each link configuration. Various simulations are conducted in order to deduce the correlation between the HMC performance and the link configuration with regard to memory utilization. It is confirmed that there is a strong correlation between the achievable maximum performance of HMC and the link configuration in terms of both bandwidth and latency. Therefore, it is possible to find the best link configuration when the required HMC performance is known in advance, and finding the best configuration will lead to significant power saving while the performance requirement is satisfied.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.6C
• /
• pp.571-580
• /
• 2002

In this paper, we relatively differentiate maximum delay for each Assured Service subclass in Differentiated Services by allocating bandwidth to each subclass differently. To maximize the throughput for the In-profile traffic and the link utilization, we propose a Cofiguration method of RIO and the admission control criterion based on the buffer size determined by the network topology and the ratio of bandwidth allocated to each subclass. Simulation results show that the proposed method can calculate the capacity to guarantee the QoS for the Assured Service and maximize the throughput for the In-profile traffic and the link utilization by applying the RIO parameter values set through the proposed configuration method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.11A
• /
• pp.1709-1716
• /
• 2000

ATM 스위치에서 입력 트래픽의 링크 이용의 불확실성과 입력 트래픽에 대한 예측 오류 등으로 인한 폭주가 발생될 가능성을 완전히 배제할 수 없기 때문에 스위치내부에서 폭주 발생을 억제할 수 있는 방법이모새되어야 한다.본 논문에서는 스위치 내부 링크의 대역 사용량을 측정하여 폭주 및 트래픽 제어를 수행하고 링크 이용률을 최대화하기 위한대역 할당 알고리즘을 제안한다. 폭주가 발생한 링크에 대하여 짧은 시간 내에 폭주를 해결하고 폭주가 발생하지 않은 링크에서는 사용하고 남은 여분의 대역을 공평하게 할당함으로서 링크의 이용률을 극대화할 수 있으므로 트래픽 관리를 보다 쉽고 효율적으로 수행할 수 있다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.19 no.7
• /
• pp.1335-1345
• /
• 1994

In this paper, the performance analysis of Inter-Processor Communication Network(IPCN) in a large-capacity digital switching system, TDX-10, is presented. The simulation model of IPCN is developed using discrete event model of SLAM II. The simulation results of maximum buffer length and mean waiting times at each node, and utilization of D-bus are derived. Finally, the maximum call handling capacity of IPCN is obtained by taking link speed into consideration.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.49 no.1
• /
• pp.35-42
• /
• 2012

This paper is about the implementation of low cost WiMAX remote radio head system with Ethernet instead of optical link. We deployed a simplified frame structure on the data frames transmitted between a basestation and remote radio heads in order to maximize Ethernet link utilization which shows relatively lower maximum throughput compared to that of optical links. In addition, a synchronization mechanism was applied on a basestation and remote radio heads placed on remote sites in order for simultaneous data transmission and reception in all remote radio heads which are essential for proper communication with terminals. These schemes are implemented with FPGA. The results of experiment with a WiMAX Femtocell show that our remote radio head systems efficiently deliver radio frames with proper timing.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.4C
• /
• pp.342-352
• /
• 2002

In this paper, we compare the performances of RIO and RIO-DC buffer management schemes for DiffServ AF PHB standardized in IETF. For the comparison, we relatively differentiate maximum delay for each Assured Service subclass in Differentiated Services by allocating bandwidth to each subclass differently. In addition, we set the values of RIO and RIO-DC parameters considering the buffer size determined by the network topology and the ratio of bandwidth allocated to each subclass. In this simulation environment, the performances of RIO and RIO-DC schemes are analyzed focusing on the throughput of the In-profile traffic, the link utilization and the fairness. Simulation results show that the performance of RIO-DC scheme is comparable to that of RIO scheme with regard to the throughput of the In-profile traffic and the link utilization. However, under the simulation condition RIO-DC scheme improves the fairness between flows much better than RIO scheme.

• Journal of Power Electronics
• /
• v.16 no.6
• /
• pp.1991-2004
• /
• 2016

An LC filter is required to reduce the output current ripple in the auxiliary resonant snubber inverter (ARSI) for high-performance applications. However, if the traditional control method is used in the ARSI with LC filter, then unnecessary current flows in the auxiliary circuit. In addressing this problem, a novel load-adaptive control that fully uses the filter inductor current ripple to realize the soft-switching of the main switches is proposed. Compared with the traditional control implemented in the ARSI with LC filter, the proposed control can reduce the required auxiliary current, contributing to higher efficiency and DC-link voltage utilization. In this study, the detailed circuit operation in the light load mode (LLM) and the heavy load mode (HLM) considering the inductor current ripple is described. The characteristics of the improved ARSI are expressed mathematically. A prototype with 200 kHz switching frequency, 80 V DC voltage, and 8 A maximum output current was developed to verify the effectiveness of the improved ARSI. The proposed ARSI was found to successfully operate in the LLM and HLM, achieving zero-voltage switching (ZVS) of the main switches and zero-current switching (ZCS) of the auxiliary switches from zero load to full load. The DC-link voltage utilization of the proposed control is 0.758, which is 0.022 higher than that of the traditional control. The peak efficiency is 91.75% at 8 A output current for the proposed control, higher than 89.73% for the traditional control. Meanwhile, the carrier harmonics is reduced from -44 dB to -66 dB through the addition of the LC filter.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.43 no.12 s.354
• /
• pp.31-41
• /
• 2006

Internet access from mobile phones over cellular networks suffer from severe bandwidth limitations and high bit error rates over wireless access links. Tailoring TCP connections to best fit the characteristics of this bottleneck link is thus very important for overall performance improvement. In this work, we propose a simple algorithm in deciding the optimal TCP segment size to maximize the utilization of the bottleneck wireless TCP connection for mobile contents server access, taking the dynamic TCP window variation into account. The proposed algorithm can be used when the product of the access rate and the propagation time is not large. With some numerical examples, it is shown that the optimal TCP segment size becomes a constant value when the TCP window size exceeds a threshold. One can set the maximum segment size of a wireless TCP connection to this optimal segment size for mobile contents server access for maximum efficiency on the expensive wireless link.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2004.05b
• /
• pp.381-385
• /
• 2004

In this paper, We consider various service models and mechanisms as a part of study for offering QoS with the requirement of user and discuss genetic algorithm and hybrid genetic algorithm for routing optimization in broa㏈and convergence network. If routing optimization based on OSPF is not sufficient, a number of MPLS paths can be set up to further improve QoS. We propose two mixed-integer programming models for the complementary MPLS problem, and consider the maximum link utilization within the network as the relevant network QoS measure

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.9
• /
• pp.1890-1906
• /
• 1997

A new ER feedback switch control algorithm, called EMRCA(Explicit Max-min Rate Control Algorithm), to control the ABR services traffic in the ATM networks, is proposed in this thesis. This scheme was interpreted the ACR variation residing on between a transient state and a stable state using the two thresholds that is the $TH_{high}$ and $TH_{low}$ in the queue to detect switch nodes congestion, and decreased the difference of minimum and maximum ACR all active connection to enhance the line bandwidth utilization. The proiposed scheme used a minimum and maximum rate of all active connections to select the connection whose the congestion rate inveoked, and uses a congestion detection mechanixm to prevent some potential cogestion by regulating selected contection.s The EMRCA scheme showed ACR variation more stable that the EPRCA scheme, also reduced buffer size of the switch, and achieved higher link utilization than the EPRCA scheme.