• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38A no.10
• /
• pp.858-864
• /
• 2013

In the 4G wireless environment, one of the important issues is to discover and select an access network suited for users. In this thesis, we propose a new network selection mechanism using group decision making and evaluate the effect of network selection schemes for vertical handover in heterogeneous wireless networks. We consider the group of users with similar QoS requirements search for the available access network simultaneously and a service area consist of multiple access networks with various characteristics. We divide the access networks with similar characteristics split into a group. Between each group, the one group is selected and within that group, the best access networks will be assigned according to priority order by network selection algorithm. We evaluate and compare the performance of three representative MADM schemes: GRA, SAW and TOPSIS. The MATLAB simulation results indicate the proposed algorithm can make a more effective choice according to the networks' characteristics and user's preference.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1627-1634
• /
• 2021

This paper introduces a low-power compact ADC circuit for analog Convolutional filter for low-power neural network accelerator SOC. While convolutional neural network accelerators can speed up the learning and inference process, they have drawback of consuming excessive power and occupying large chip area due to large number of multiply-and-accumulate operators when implemented in complex digital circuits. To overcome these drawbacks, we implemented an analog convolutional filter that consists of an analog multiply-and-accumulate arithmetic circuit along with an ADC. This paper is focused on the design optimization of a low-power 8bit SAR ADC for the analog convolutional filter accelerator We demonstrate how to minimize the capacitor-array DAC, an important component of SAR ADC, which is three times smaller than the conventional circuit. The proposed ADC has been fabricated in CMOS 65nm process. It achieves an overall size of 1355.7㎛², power consumption of 2.6㎼ at a frequency of 100MHz, SNDR of 44.19 dB, and ENOB of 7.04bit.

• Proceedings of the Korean Information Science Society Conference
• /
• 2010.11a
• /
• pp.101-106
• /
• 2010

As a part of preparing the sniffing attack, this routing method presented in this thesis decreases the risk rates of the leaking of information through separating valid data and transmitting by a multi-path. then data is transmitted from start node to destination node on distributed sensor network. The level of reduction in leaking of information by the sniffing attack is proved by experimental result thich compare the case described above with the case of transmitting whole data with the single path by simulation, and the algorithm for choosing the routing path is showed by proof of the theorem.

• Journal of Korean Society of Transportation
• /
• v.23 no.7 s.85
• /
• pp.111-123
• /
• 2005

A network design model has been proposed. which represents a transportation facility investment decision problem. The model takes the discrete hi-level programming form in which two types of decision makers, government and travelers, are involved. The model is characterized by its ability to address the total social costs occurring in transportation networks and to estimate the equilibrium link volumes in multi-modal networks. Travel time and volume for each link in the multi-modal network are predicted by a joint modal split/traffic assignment model. An efficient solution algorithm has been developed and an illustrative example has been presented.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.9
• /
• pp.4301-4319
• /
• 2017

Adopting mobile chargers (MC) in rechargeable wireless sensors network (R-WSN) to recharge sensors can increase network efficiency (e.g., reduce MC travel distance per tour, reduce MC effort, and prolong WSN lifetime). In this study, we propose a mechanism to split the sensing field into partitions that may be equally spaced but differ in distance to the base station. Moreover, we focus on minimizing the MC effort by providing a new charging mechanism called the sector-based charging schedule (SBCS), which works to dispatch the MC in charging trips to the sector that sends many charging requests and suggesting an efficient sensor-charging algorithm. Specifically, we first utilize the high ability of the BS to divide the R-WSN field into sectors then it select the cluster head for each sector to reduce the intra-node communication. Second, we formulate the charging productivity as NP-hard problem and then conduct experimental simulations to evaluate the performance of the proposed mechanism. An extensive comparison is performed with other mechanisms. Experimental results demonstrate that the SBCS mechanism can prolong the lifetime of R-WSNs by increasing the charging productivity about 20% and reducing the MC effort by about 30%.

• Journal of Communications and Networks
• /
• v.4 no.2
• /
• pp.108-117
• /
• 2002

TCP, which was developed on the basis of wired links, supposes that packet losses are caused by network congestion. In a wireless network, however, packet losses due to data corruption occur frequently. Since TCP does not distinguish loss types, it applies its congestion control mechanism to non-congestion losses as well as congestion losses. As a result, the throughput of TCP is degraded. To solve this problem of TCP over wireless links, previous researches, such as split-connection and end-to-end schemes, tried to distinguish the loss types and applied the congestion control to only congestion losses; yet they do nothing for non-congestion losses. We propose a novel transport protocol for wireless networks. The protocol called VS-TCP (Variable Segment size Transmission Control Protocol) has a reaction mechanism for a non-congestion loss. VS-TCP varies a segment size according to a non-congestion loss rate, and therefore enhances the performance. If packet losses due to data corruption occur frequently, VS-TCP decreases a segment size in order to reduce both the retransmission overhead and packet corruption probability. If packets are rarely lost, it increases the size so as to lower the header overhead. Via simulations, we compared VS-TCP and other schemes. Our results show that the segment-size variation mechanism of VS-TCP achieves a substantial performance enhancement.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.36 no.7A
• /
• pp.669-677
• /
• 2011

Superframe in IEEE 802.15.4 Standard is subdivided into an active period and an inactive period to reduce energy consumption. But communication nodes use same data transmission range in an active period, thus communication nodes spend a lot of energy to send data another nodes. In this paper, we proposed reduce energy consumption algorithm that nodes use different transmission power. Cordinator split transmission area into four group and transmit becon message to nodes. Nodes adjust transmission power according to becon message and wates lowe energy than normal nodes. We proposed energy-efficient cross layer protocol that have different PAN (Personal Area Network) by four range group.

• Journal of Broadcast Engineering
• /
• v.10 no.1 s.26
• /
• pp.83-102
• /
• 2005

The problem of video scheduling is analyzed in the framework of divisible load scheduling. A divisible load can be divided into any number of fractions (parts) and can be processed/computed independently on the processors in a distributed computing system/network, as there are no precedence relationships. In the video scheduling, a frame can be split into any number of fractions (tiles) and can be processed independently on the processors in the network, and then the results are collected to recompose the single processed frame. The divisible load arrives at one of the processors in the network (root processor) and the results of the computation are collected and stored in the same processor. In this problem communication delay plays an important role. Communication delay is the time to send/distribute the load fractions to other processors in the network. and the time to collect the results of computation from other processors by the root processors. The objective in this scheduling problem is that of obtaining the load fractions assigned to each processor in the network such that the processing time of the entire load is a minimum. We derive closed-form expression for the processing time by taking Into consideration the communication delay in the load distribution process and the communication delay In the result collection process. Using this closed-form expression, we also obtain the optimal number of processors that are required to solve this scheduling problem. This scheduling problem is formulated as a linear pro-gramming problem and its solution using neural network is also presented. Numerical examples are presented for ease of understanding.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.63 no.5
• /
• pp.713-718
• /
• 2014

DC urban railway power system consists of DC power network and AC power network. The DC power network supplies electric power to railway vehicles and the AC power network supplies electric power to station electric equipment. Recently, because of power consumption reduction and peak load shaving, intelligent measurement of regenerative energy and renewable energy adapted on DC urban railway is required. For this reason, DC smart metering system for DC power network shall be developed. Therefore, in this paper, DC voltage sensor, current sensor, and DC smart meter were developed and evaluated by performance test. DC voltage sensor was developed for measuring standard voltage range of DC urban railway, and DC current sensor was developed as hall effect split core type in order to install in existing system. DC smart meter possesses function of general intelligent electric power meter, such as measuring electricity and wireless communication etc. And, DC voltage sensor showed average 0.17% of measuring error for 2,000V/50mA, and current sensor showed average 0.21% of measuring error for ${\pm}2,000V/{\pm}4V$ in performance test. Also DC smart meter showed maximum 0.92% of measuring error for output of voltage sensor and current sensor. In similar environment for real DC power network, measuring error rate was under 0.5%. In conclusion, accuracy of DC smart metering system was confirmed by performance test, and more detailed performance will be verified by further real operation DC urban railway line test.

• Imaging Science in Dentistry
• /
• v.52 no.2
• /
• pp.219-224
• /
• 2022

Purpose: This study aimed to evaluate the performance of transfer learning in a deep convolutional neural network for classifying implant fixtures. Materials and Methods: Periapical radiographs of implant fixtures obtained using the Superline (Dentium Co. Ltd., Seoul, Korea), TS III(Osstem Implant Co. Ltd., Seoul, Korea), and Bone Level Implant(Institut Straumann AG, Basel, Switzerland) systems were selected from patients who underwent dental implant treatment. All 355 implant fixtures comprised the total dataset and were annotated with the name of the system. The total dataset was split into a training dataset and a test dataset at a ratio of 8 to 2, respectively. YOLOv3 (You Only Look Once version 3, available at https://pjreddie.com/darknet/yolo/), a deep convolutional neural network that has been pretrained with a large image dataset of objects, was used to train the model to classify fixtures in periapical images, in a process called transfer learning. This network was trained with the training dataset for 100, 200, and 300 epochs. Using the test dataset, the performance of the network was evaluated in terms of sensitivity, specificity, and accuracy. Results: When YOLOv3 was trained for 200 epochs, the sensitivity, specificity, accuracy, and confidence score were the highest for all systems, with overall results of 94.4%, 97.9%, 96.7%, and 0.75, respectively. The network showed the best performance in classifying Bone Level Implant fixtures, with 100.0% sensitivity, specificity, and accuracy. Conclusion: Through transfer learning, high performance could be achieved with YOLOv3, even using a small amount of data.