Earticle

There is an increasing reliance on wireless computer networks for communicating various types of time sensitive applications such as voice over internet protocol (VoIP). Quality of service (QoS) can play an important role in wireless computer networks as it can facilitate evaluation of their performance and can provide mechanisms to improve their operation. In this study probabilistic neural network (PNN) and Bayesian classification were developed to process delay, jitter and percentage packet loss ratio for VoIP traffic. Both methods successfully categorized the transmission of VoIP packets into low, medium and high QoS categories but overall the Bayesian approach performed more accurately than PNN. By accurately determining the network’s QoS, an improved understanding of its performance is obtained.

정보의 양이 급격히 증가함에 따라 정보에 대한 관리 및 보안 취약점들이 다 양하게 존재하게 되었다. 뿐만 아니라 대학의 경우 다수의 IT자원에 대비하여 관 리 인력이 부족함에 따라서 외부의 위협에 대처할 수 있는 방안이 부족한 실정 이다. 따라서 본 연구에서는 보안 에이전트 도입을 통해서 소수의 보안 담당자 만으로도 효과적으로 보안 관제를 수행할 수 있는 환경을 마련한다. 또, 좀비 PC의 발생에 대처하기 위한 좀비PC탐지 기법과 외부 침입 탐지 및 방지 기술을 조합하여 대학 환경에 적합한 보안관제 방안을 제시한다. 주제어: 보안관제,

With the rapid increase of information, the security vulnerability are emerging issue in information management. In addition, coverage of security monitoring services in university is so large, but they have the problem of the lack of administrative staff. This paper suggests the appropriate solutions against external cyber-attacks. They are anti-virus solutions, detections of zombie PC and the intruder detections and preventions.

신경망은 데이터로부터 반복적인 학습 과정을 통해 숨어 있는 패턴을 찾아내고, 새로운 데이터의 목표값에 대한 정확한 예측에 유용한 모델링 기법이다. 본 논문은 개인적인 특성, 가정․사회적 환경, 타 교과 성적을 이용하여 학생의 컴퓨터 활용 능력 예측을 위한 다층 인식모형(MLP) 신경망을 구축하였다. 신경망의 인식률은 예측 방법으로 널리 활용되고 있는 로지스틱 회귀분석 모델과 비교하였다. 개발한 신경망에 대한 실험 결과, 개인적인 특성이 학생들의 컴퓨터 활용 능력을 가장 잘 설명하는 요소이며, 반면 가정․사회적 환경은 가장 낮은 예측 요소임을 발견하였다. 또한 본 연구의 신경망 모델은 회귀분석보다 더욱 높은 인식률을 나타냈다.

A neural network is a modeling technique useful for finding out hidden patterns from data through repetitive learning process and for predicting target values for new data. In this study, we built multilayer perceptron neural networks for prediction of the students' computer literacy based on their personal characteristics, home and social environment, and academic record of other subjects. Prediction performance of the network was compared with that of a widely used prediction method, the regression model. From our experiments, it was found that personal characteristic features best explained computer proficiency level of a student, whereas the features of home and social environment resulted in the worse prediction accuracy among all. Moreover, the developed neural network model produced far more accurate prediction than the regression model.

Distributed Systems remained one of the most recent development fields in computing research. The parallel applications development faced enormous hindrance in QoS delivery due to resource hungry queuing models and complex memory patterns. Being a collection of multiple autonomous systems, a distributed system's efficiency lies in best coordination of its middle ware and coordinating modular layers. The concurrency is maintained by implementing Integrity constraint rules. The paper assesses various coherence addressing schemes for QoS delivery in Distributed Network environment.

Available bandwidth is one the key challenges that is present in every new developed standard in the computer networks. There are many proposed methodologies in the existing research literature for measuring the available bandwidth. In this paper we will present summary of the main contributions in this field, research problems and solutions and we will present possible ways for improving the suggested solutions in the existing literature. Furthermore, survey of the most important tools for measuring the available bandwidth in wired and wireless networks is presented in details.

Many various kinds of network applications have arisen due to rapid development of network techniques. For preventing sensitive personal information from being disclosed on an open and unsecure network, it is necessary to provide some appropriate secure protocols. Concerning secure protocols, key agreement and authentication between user and server are paramount. In 2010, Cui and Cao proposed a secure anonymous key agreement for distributed networks, in which users collect other identities to utilize in communication so attackers cannot determine the real identity of the user. However, this protocol suffers the drawback of high calculation requirements. In this paper, we adopt Elliptic Curve Cryptography (ECC) to reduce the computational cost in Cui and Cao’s protocol, and propose an indexing trick to speed up searches of legitimate users. Our proposed scheme maintains the characteristic of obfuscating user identity to thwart identification attempts.

The field of WLAN has grown significantly in the last three decades. An interesting usage of computer networks is in offices and educational institutes were increased to exchange streams of multimedia data. These privately-owned networks are known as Local Area Networks (LANs) which come under the category of small scale networks but researchers introduced the possible usage of wireless technologies for interconnection. This has resulted in the emergence of wireless LANs (WLANs). In the world of information system, performance is a very important subject. It consists of the work which the computer WLAN system is supposed to do, and how do we judge whether is being done well. To achieve performance in the field of WLAN, certain features must be available and operational in the design of the WLANs. Performance must also be considered when configuring and implementing the WLAN systems. The modeler of such WLAN systems must follow a systematic approach in evaluation, measurements, and analytical or simulation modeling of such WLAN system. This paper presents a comparison and an evaluation of transmitting multimedia streams over three different WLANs scenarios by using OPNET simulator

Considering that intrusion detection systems and anomaly clearly recognize malicious activity. Nowadays, data mining based intrusion detection systems, security and more rapidly detect attacks.Therefore,in this article we use a combination of k-means clustering algorithm and is used supervised support vector machine algorithm to find the best line separator. This is leading to the separation of normal and attack traffic.

In recent years, using computer networks (wired and wireless networks) has been widespread in many applications. As computer networks become increasingly complex, the accompanied potential threats also grow to be more sophisticated and as such security has become one of the major concerns in them. Prevention methods alone are not sufficient to make them secure; therefore, detection should be added as another defense before an attacker can breach the system. Intrusion Detection Systems (IDSs) have become a key component in ensuring systems and networks security. An IDS monitors network activities in order to detect malicious actions performed by intruders and then initiate the appropriate countermeasures. In this paper, we present a survey and taxonomy of intrusion detection systems and then evaluate and compare them.

A terminal response time analysis for a general class of terminals-to-computer subsystem is presented in this paper. On the point of the front view, it should be considered for R.O.K. Military Defense to set up the communication network in order to facilitate for the currency of the information and the data communication system. The model used to study is based on the advanced data communications system in which terminals are connected to Terminal Control Units(TCU) that are in turn connected to local Front-End Processor(FEP). The line control procedures used to interface a TCU and an FEP may be half-duplex Binary Synchronous Communication(BSC), half-duplex Synchronous Data Link Control(SDLC), or full-duplex SLDC. This paper will contribute to facilitate the initial phase of system design and configuration for the Military Defense Communication Network System in future.

An optimal routing method using hysteresis binary neurons with link failure probability is proposed in this paper. The link failures in computer networks can degrade the performance of the entire networks. We assume the time between successive link failures is exponentially distributed with parameter ${\lambda}$ and the failures are independent. The link failure probability is used for neural networks to find the shortest paths of given source-destination pairs. By using the probability of link failures and hysteresis binary neurons we implement an optimal routing method that can takes routes by coping with link failures.

This paper presents an application of heuristic approach to problem of designing reliable network expansion. The problem essentially consists in finding the network topology that satisfies given set of reliability constraints. To efficiently solve the problem, a scatter search approach is proposed. The results of the two experiments show that scatter search is a more suitable approach for finding a good solution or near optimal solution in comparison with genetic algorithm.

This paper dealt with the development of a surge protection device (SPD) that can protect high speed computer network devices from overvoltages caused by switching operations or lightning surges. The designed SPD is a form of hybrid circuit which is composed of a gas tube having large current diverting capability, high response bi-directional avalanche diodes, and fast recovery diodes to reduce insertion loss on high frequency domain. Surge protection and signal transmission characteristics of the fabricated SPD was tested according to the international standards, IEC 61000-4-5 and IEC 61643-21. From the test results, the SPD is satisfied with the international standards and the high cut-off frequency was 204MHz. Also, the SPD showed a good performance without an insertion loss on a field test of 100Mbps class Local Area Network

통신망의 규모가 커지고 구조가 복잡해짐에 따라, 통신망의 성능을 최적화하여 사용자들이 요구하는 서비스 품질을 보장해 주는 성능관리의 기능이 절실히 요구되고 있다. 현재 성능관리의 주요 기능이 되는 트래픽에 대한 분석을 위해서 넷플로우(NetFlow), RMON, 그리고 패킷을 캡처하는 방법이 쓰이고 있지만 통합적인 관점의 해결책은 되지 못한다. 본 논문에서는 다양한 전송기술(Multi-technology), 다양한 장치 제조사(Multi-vender) 장비들의 성능관리를 가능케 할 수 있도록 통합적인 통신망 성능관리 구조를 제시하고, 그에 따라 성능관리 시스템을 설계하고 구현하였다.

인터넷을 중심으로 글로벌네트워크화 되고 있는 국내 전산망에서는 국내외 해킹 등에 의한 전산망 보안 침해사고가 매우 우려될 뿐 아니라 빈번히 발생되고 있다. 하지만 이에 대한 대책과 기술이 제대로 이루어지지 않고 있으며, 해외의 전문 침해사고대응에 비해 매우 열악한 상태에 있으므로 침해사고의 예방, 사고시 분석 및 확산 방지 및 방지책 개발과 이의 상호 정보 공유 방법을 제시하고 운영하고자 한다.

본 논문은 이 세상에 존재하는 어떤 종류의 보안시스템도 시스템 접근권한을 가진 전산 담당자의 시스템 침해 행위를 막아내기란 어려운 일이라는 점을 되새기고, 전산 담당자의 침해 행위 사례와 그로인한 피해의 정도 등 심각성을 재조명하고, 나아가 기술적 보안조치의 한계선상에 있는 전산 담당자의 권한 관리와 기술적 조치를 넘어 인적관리를 통한 침해예방 방안에 대해 나름의 대안을 찾아 보고자 한다.

In this paper, to remind ourselves that exist in the world what kind of security system is also difficult to defend a system with computerized access system representative of infringement that, Computerized personnel act of infringement cases, and thus the extent of the damage and the severity Revisited, Furthermore, technical measures beyond the limits of the technical security measures in line computerized rights management representative and for infringement prevention measures in Human Resource Management through its own alternative to find.

본 논문은 금융투자업계에 대한 보안관제 정책을 기준으로 업계가 공동이용 가능한 통합보안관제시스템이 구축 가능하도록 금융투자회사들이 구축 운용하고 있는 보안인프라를 비롯한 전산망의 보안관제를 위한 적용 기술과 운용체계 등을 조사하고, 이를 토대로 외부로부터 공격에 대비한 모니터링, 침입탐지 및 실시간 방어 등의 기능이 적절하게 수행되고 업계 차원의 종합적이고 체계적인 관리가 가능한 통합보안관제시스템 모델과 운영방안을 제시하고자 한다.

본 논문은 금융투자업계에 대한 보안관제 정책을 기준으로 업계가 공동이용 가능한 통합보안관제시스템이 구축 가능하도록 금융투자회사들이 구축 운용하고 있는 보안인프라를 비롯한 전산망의 보안관제를 위한 적용 기술과 운용체계 등을 조사하고, 이를 토대로 외부로부터 공격에 대비한 모니터링, 침입탐지 및 실시간 방어 등의 기능이 적절하게 수행되고 업계 차원의 종합적이고 체계적인 관리가 가능한 통합보안관제시스템 모델과 운영방안을 제시하고자 한다.