• Journal of Electrical Engineering and Technology
• /
• v.3 no.2
• /
• pp.285-291
• /
• 2008

Active Queue Management(AQM) has been widely used for congestion avoidance in Transmission Control Protocol(TCP) networks. Although numerous AQM schemes have been proposed to regulate a queue size close to a reference level, most of them are incapable of adequately adapting to TCP network dynamics due to TCP's non-linearity and time-varying stochastic properties. To alleviate these problems, we introduce an AQM technique based on a dynamic neural network using the Back-Propagation(BP) algorithm. The dynamic neural network is designed to perform as a robust adaptive feedback controller for TCP dynamics after an adequate training period. We evaluate the performances of the proposed neural network AQM approach using simulation experiments. The proposed approach yields superior performance with faster transient time, larger throughput, and higher link utilization compared to two existing schemes: Random Early Detection(RED) and Proportional-Integral(PI)-based AQM. The neural AQM outperformed PI control and RED, especially in transient state and TCP dynamics variation.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.6A
• /
• pp.910-917
• /
• 2000

Because the random early detection(RED) algorithm deals all flows with the same best-effort traffic characteristic, it can not correctly control the output link bandwidth for the flows with different traffic characteristics. To remedy this problem, several per-flow algorithms have been proposed. In this paper, we propose a new per-flow type Fair Droptail algorithm which can fairly allocate bandwidth among flows over a shared output link. By evenly allocating buffers per flow, the Fair Droptail can restrict a flow not to use more bandwidth than others. In addition, it can be simply implemented even if it employs the per-flow state mechanism, because the Fair Droptail only keeps each information of flow in active state.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.11a
• /
• pp.354-357
• /
• 2020

DDoS(Distributed Denial of Service) 공격은 현재의 인터넷 환경뿐만 아니라 NDN에서도 정상적인 서비스를 저해시키는 주요 문제이며 이에 관련된 다양한 연구들이 진행되고 있다. 본 논문에서는 DDoS 공격이 가해질 때 NDN 라우터의 PIT(Pending Interest Table) 가용성 저해로 인해 발생하는 문제 해결에 중점을 둔다. 이를 위한 방안으로 RED(Random Early Detection) 알고리즘을 기반으로 하는 기법을 적용하고, 시뮬레이션을 통한 측정 결과를 보여준다.

• International Journal of Computer Science & Network Security
• /
• v.22 no.2
• /
• pp.123-130
• /
• 2022

During the past decades, detection of gear defects remains as a major problem, especially when the gears are subject to non-stationary phenomena. The idea of this paper is to mixture a multilevel wavelet transform with a fast EMD decomposition in order to early detect gear defects. The sensitivity of a kurtosis is used as an indicator of gears defect burn. When the gear is damaged, the appearance of a crack on the gear tooth disrupts the signal. This is due to the presence of periodic pulses. Nevertheless, the existence of background noise induced by the random excitation can have an impact on the values of these temporal indicators. The denoising of these signals by multilevel wavelet transform improves the sensitivity of these indicators and increases the reliability of the investigation. Finally, a defect diagnosis result can be obtained after the fast transformation of the EMD. The proposed approach consists in applying a multi-resolution wavelet analysis with variable decomposition levels related to the severity of gear faults, then a fast EMD is used to early detect faults. The proposed mixed methods are evaluated on vibratory signals from the test bench, CETIM. The obtained results have shown the occurrence of a teeth defect on gear on the 5th and 8th day. This result agrees with the report of the appraisal made on this gear system.

• Journal of Korea Society of Industrial Information Systems
• /
• v.28 no.5
• /
• pp.109-126
• /
• 2023

Early detection of forest fires is essential in preventing large-scale forest fires. Predicting forest fires serves as a vital early detection method, leading to various related studies. However, many previous studies focused solely on climate and geographic factors, overlooking human factors, which significantly contribute to forest fires. This study aims to develop forest fire prediction models that take into account human, weather and geographical factors. This study conducted a comparative analysis of four machine learning models alongside the logistic regression model, using forest fire data from Gangwon-do spanning 2003 to 2020. The results indicate that XG Boost models performed the best (AUC=0.925), closely followed by Random Forest (AUC=0.920), both of which are machine learning techniques. Lastly, the study analyzed the relative importance of various factors through permutation feature importance analysis to derive operational insights. While meteorological factors showed a greater impact compared to human factors, various human factors were also found to be significant.

• Physical Therapy Korea
• /
• v.7 no.3
• /
• pp.1-18
• /
• 2000

The present study was performed to investigate the prevalence rate of idiopathic scoliosis and to determine the effect of exercise training on scoliotic angle in elementary school children. In this study, two out of five elementary schools in Seosan city were chosen by random sampling. Seven hundred sixty four students (from four grade to the sixth grade student) were selected in two schools. Screening tests were conducted to find idiopathic scoliosis. Among the 764 individuals, 139 subjects who showed positive sign in physical examination took whole spine radiography. Thirty six subjects who had a curve of 10 or greater and consented to participate in the exercise program were selected for the exercise program. The exercise program was performed four times a week for 5 months. The results of this study were as follows: 1) One hundred thirty nine subjects showed positive sign in the scoliosis screening test. 2) The overall prevalence of curve of $10^{\circ}or$ greater in X-ray finding was 8.15%. The prevalencies of curve of $10^{\circ}or$ greater in male and female were 7.1% and 9.2%, respectively. 3) Scoliosis curves were observed at thoracic area (48.4%), at thoracolumbar area (27.4%) and at lumbar area(24.4%). 4) Right side curve was 59.7%, and left side curve was 40.3%. 5) After the 5 month exercise program for scoliosis, the Cobb's angle was significantly decreased. 6) There was no significant difference of Cobb's angle change respect to sex, grades, and scoliosis curve site. Results shown here indicates that an early detection and early exercise for scoliosis can result in decreased the Cobb's angle in elementary school children.

• Journal of Information Processing Systems
• /
• v.5 no.1
• /
• pp.33-40
• /
• 2009

Ever since the network-based malicious code commonly known as a 'worm' surfaced in the early part of the 1980's, its prevalence has grown more and more. The RCS (Random Constant Spreading) worm has become a dominant, malicious virus in recent computer networking circles. The worm retards the availability of an overall network by exhausting resources such as CPU capacity, network peripherals and transfer bandwidth, causing damage to an uninfected system as well as an infected system. The generation and spreading cycle of these worms progress rapidly. The existing studies to counter malicious code have studied the Microscopic Model for detecting worm generation based on some specific pattern or sign of attack, thus preventing its spread by countering the worm directly on detection. However, due to zero-day threat actualization, rapid spreading of the RCS worm and reduction of survival time, securing a security model to ensure the survivability of the network became an urgent problem that the existing solution-oriented security measures did not address. This paper analyzes the recently studied efficient dynamic network. Essentially, this paper suggests a model that dynamically controls the RCS worm using the characteristics of Power-Law and depth distribution of the delivery node, which is commonly seen in preferential growth networks. Moreover, we suggest a model that dynamically controls the spread of the worm using information about the depth distribution of delivery. We also verified via simulation that the load for each node was minimized at an optimal depth to effectively restrain the spread of the worm.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.4
• /
• pp.2429-2432
• /
• 2013

Background and Objectives: Yanting in Sichuan Province is one of the highest risk areas of esophageal cancer (EC) in the world. We here summarize the epidemiology of EC in Yanting using data from the national screening programme during 2006-2011. Methods: Random cluster sampling was used to select a proportion of natural villages from six towns in Yanting, and residents aged 40-69 years old were invited for screening. Participants were screened using endoscopy with iodine staining and then confirmed by histological examinations. Results: The overall detection rates of low-grade hyperplasia (LH), moderate hyperplasia (MH), high-grade hyperplasia (HH), carcinoma in situ (CIS), intramucosal carcinoma (IC) and invasive carcinoma (INC) were 5.33%, 1.28%, 0.68%, 0.15%, 0.06% and 0.29%, respectively. The detection rates of LH, MH, HH and INC increased with age, reaching the peak among those aged 60-65 years, and the prevalences of LH and MH were higher among men than among women. In addition, the detection rates of hyperplasia were much higher in mountainous than in hilly areas. Conclusions: Among the high risk population, there are a great number of people with early-stage EC or precancerous conditions who do not have presenting symptoms. In particular, the elderly, men, or those living in mountainous areas are the most vulnerable population. It is therefore important to reinforce health education and screening services among such high risk populations.

• Journal of Global Scholars of Marketing Science
• /
• v.16 no.1
• /
• pp.13-26
• /
• 2006

In this research, we develope a procedure for detecting a random non-stationarity to the individual's purchasing rate in a stationary NED model. On this purpose, we derive the likelihood ratio statistic for a testing null and alternative hypotheses defined as whether there is no significant structural change in a stationary NED model or any. Where the structural change comes from a random non-stationarity(marketing mix activities or seasonality, for example) to the individual's purchasing rate. We also apply the developed method to a panel data for a frequently purchased good. This research could be a solution to include the non-stationarity in a stationary NED model. We also expect that the developed model could give a signal for an early detection of significant changes in marketing environment, and a mean for a measurement of the effects of marketing mix activities.

• Journal of Internet Computing and Services
• /
• v.23 no.3
• /
• pp.35-45
• /
• 2022

Even in recent years, treatment of first-aid patients is still often delayed due to a shortage of medical resources in marginalized areas. Research on automating the analysis of medical data to solve the problems of inaccessibility for medical services and shortage of medical personnel is ongoing. Computer vision-based medical inspection automation requires a lot of cost in data collection and labeling for training purposes. These problems stand out in the works of classifying lesion that are rare, or pathological features and pathogenesis that are difficult to clearly define visually. Anomaly detection is attracting as a method that can significantly reduce the cost of data collection by adopting an unsupervised learning strategy. In this paper, we propose methods for detecting abnormal images on chest X-RAY images as follows based on existing anomaly detection techniques. (1) Normalize the brightness range of medical images resampled as optimal resolution. (2) Some feature vectors with high representative power are selected in set of patch features extracted as intermediate-level from lesion-free images. (3) Measure the difference from the feature vectors of lesion-free data selected based on the nearest neighbor search algorithm. The proposed system can simultaneously perform anomaly classification and localization for each image. In this paper, the anomaly detection performance of the proposed system for chest X-RAY images of PA projection is measured and presented by detailed conditions. We demonstrate effect of anomaly detection for medical images by showing 0.705 classification AUROC for random subset extracted from the PadChest dataset. The proposed system can be usefully used to improve the clinical diagnosis workflow of medical institutions, and can effectively support early diagnosis in medically poor area.