Defending denial-of-service attacks in SDNFV-enabled cloud computing

Phan, Van Trung Soongsil University 2017 국내석사

In my thesis, I propose a novel S2eH scheme to tackle Denial-of-Service (DoS) attacks in the Software-Defined Network Functions Virtualization (SDNFV) cloud computing environment. I firstly introduce a new machine learning hybrid model for DoS attack classification based on Support Vector Machine (SVM) and Self Organizing Map (SOM) algorithms to enhance the performance of classification network traffic. The proposed combination mainly focuses on taking advantages of two classification algorithms by utilizing both algorithm advantages that SVM takes a little time to produce outputs with a high accuracy and SOM makes a reliable prediction based on their neurons. Then, I propose an enhanced History-based IP Filtering scheme (eHIPF) to improve attack detection rate and speed. Finally, I propose a novel mechanism combining both the machine learning hybrid model SVMs-SOM and the eHIPF scheme, called S2eH, to make a DoS attack defender in the SDNFV-enabled cloud computing. The S2eH testbed is implemented in the SDNFV cloud environment with Service Function Chaining. Through practical experiments in this testbed, it is proved that the proposed SVMs-SOM combination and eHIPF scheme outperforms existing mechanisms for DoS attack classification and detection. By analyzing comprehensive experiments conducted with various DoS attack levels, I prove that the novel S2eH mechanism is an effective and innovative approach to defend DoS attacks in the SDNFV-based cloud computing. 본 논문에서는 Software-Defined Network Functions Virtualization (SDNFV) Cloud computing 환경에서 서비스 거부 공격을 저지하는 새로운 S2eH 스키마를 제안한다. 먼저, DoS 공격 분류를 위해 Network 분류 성능을 높이는 Support Vector Machine (SVM) 과 Self-Organizing Map(SOM) 알고리즘 기반 새로운 Hybrid Machine Learning을 소개한다. 제안된 결합 시스템은 주로 두 가지 분류 알고리즘의 이점을 가져오는데 초점을 둔다. SVM은 짧은 시간에 높은 정확도로 분류를 하고, SOM은 SOM의 신경망기반으로 신뢰성이 높은 예측을 한다. 또한, 공격 탐지 비율과 속도를 개선하기 위해 History 기반 IP Filtering 스키마(eHIPF)를 제안한다. 본 논문에서는 SDNFV가 가능한 Cloud Computing 에서 DoS공격을 방어하기 위해 SVMs-SOM을 결합한 Hybrid Machine Learning 모델과 eHIPF 스키마를 결합한 새로운 메커니즘 S2eH을 제안한다. S2eH testbed는 Service Function Chaining이 가능한 SDNFV Cloud 환경에서 실험했다. 이 testbed에서 실제 실험을 통해 본 논문에서 제안된 SVMs-SOM 결합과 eHIPF 스키마가 DoS공격 분류와 탐지에 다른 메커니즘보다 좋은 결과 가져오는 것을 증명했다. 다양한 DoS공격 레벨로 수행 된 포괄적인 실험을 분석하여 새로운 S2eH메커니즘이 SDNFV기반 Cloud computing에서 DoS 공격을 방어하기 위한 효과적이고 혁신적인 방법임을 증명했다.

Expanding constraint theory to determine well-posedness of large mathematical models

Phan, Phan University of Southern California 2011 해외박사(DDOD)

Mathematical modeling represents one of the major tools for the conception and management of the ever increasing complexity of systems engineering. Unfortunately, present approaches to math modeling suffer from several theoretical problems which include: model consistency, computational allowability, management of the topologically complex flow of software algorithms, rearrangement of independent and dependent variables, distinction between the model structure and software programs and perhaps the most challenging, the exponential explosions resulting from the management of even medium sized models as well as the large models of the thousands of dimensions necessary to define and manage the complex systems of the future. Constraint theory was designed to solve the above problems employing a rigorous application of graph theory and attempts to employ the generalizability of mathematics to extend the math model manager's conceptual understanding from half a dozen dimensions to the desired thousands of dimensions. Constraint theory (CT) went through several stages of detail and maturity, starting with the PhD dissertation in the 1960's and progressing through several papers and two other PhD research programs. CT's present frontier can be characterized by the Constraint Theory book written by Dr. George Friedman and published by Springer in 2005. CT differs from linear programming (LP) in several ways. LP requires a full model of explicit mathematical expressions in linear form whereas CT employs a meta-model based upon relevancy between general (linear or non-linear) relations and variables. CT seeks to determine a model consistency and computational allowability without having to actually solve for a specific solution set whereas LP assumes that the problem is well-posed while attempting to solve for an optimal solution within a given constrained trade space. As mentioned above, the exponential explosions associated with the management of thousand-dimensional models is truly enormous, on the order of 2 N examinations of a model's N equations are required to determine consistency, for example. CT has converged this process---which would take several universe lifetimes even with nanosecond computer cycle times---by a factor of trillions. This convergence is based on an ordered series of graph theoretic steps involving connected-ness, tree-ness, circuit-ness and cluster-ness. A constraint theoretic structure called the "Basic Nodal Square" (BNS) is identified as the kernel of constraint in a math model. An n x n BNS is essentially a complete sub-system of n relations and n variables within the overall model. However, Friedman's book effectively stops at the identification of circuit clusters of approximately 30 relations, and suggests that BNS within these clusters can be found by modern computers in a few hours. It is suggested that perhaps more research can employ the topological property of adjacency to converge the search for BNS within larger circuit clusters. The central contribution resultant from this investigation realizes and improves the computational efficiency of BNS search by factors of trillions, asymptotically (see Figure 10-2). This improvement has been accomplished by innovative application of graph theory, topology, algorithm analysis and linear algebra, which were not addressed to sufficient depth in the original efforts. Leveraging research results in graph theory since the early 1970's, several BNS search methods, based on nodal adjacency, circuit adjacency and nodal degree, have been developed and compared against the baseline (brute-force) approach of 2N. Primary key research findings and enablers include: (a) Decomposition of a model graph into its connected components by employing the graph-theoretic concept of a spanning tree and applying the depth-first search algorithm. (b) Innovation of the edge-centric method, over the legacy approach of vertex-centricity, to identify, and remove, internal trees (or bridges) within a connected component. (c) Further isolation of circuit clusters containing potential BNS by using the graph-theoretic concept of articulation point (or separating vertex). (d) Rigorous proof of circuit vector-based theorems to simplify the computational complexity of constructing unions of adjacent circuits, and thus reduce BNS search space. (e) Application of vectorial dot product to detect adjacency among circuits, and overlapping among nodes or BNSs. (f) Development and demonstration of a meta-meta-model graph to represent overlapping among nodes and to reduce the solution-time for BNS search, from exponential to polynomial. Additional accomplishments further improve the utility of CT by developing an integrated set of efficient computing algorithms to determine model consistency and computational allowability. The output of these algorithms can also advise the model builder of repair alternatives to correct any model inconsistency detected. Such algorithms are necessary to bridge the gap between theoretical abstracts and practical realization of CT in terms of an effective computer-assisted tool for math model management.

Research on insurance fraud and regulations on prevention of automobile insurance fraud : experiences from the United States of America and proposals for Vietnam

Phan, Thi Huong Giang TLBU Graduate School of Law 2019 국내석사

링 마우스 : 손가락 움직임에 의한 컴퓨터 무선 입력 장치

Phan, Ke Hien 한국산업기술대학교 일반대학원 2018 국내석사

국문요약 링 마우스: 손가락 움직임에 의한 컴퓨터 무선 입력 장치 한국산업기술대학교 일반대학원 전자공학과: 판게히엔 지도교수: 정 두 희 이 논문에서는 링과 같이 손가락으로 감싸는 컴퓨터 용 무선 장치를 연구했습니다. Ring Mouse는 Bluetooth를 통해 컴퓨터에 무선으로 연결된 마우스 역할을하는 스마트 장치입니다. 손가락 움직임으로 컴퓨터의 마우스 커서를 제어 할 수 있습니다. 마이크로 컨트롤러 Cortex M4와 함께 nRF52832를 사용함으로써 우리는 전력 효율이 높고 비용 효율적인 회로를 만들었습니다. 모션 데이터는 MPU-9250 모듈을 사용하여 획득합니다. 장치의 기계적 케이스는 3D 인쇄 기술로 설계 및 제작되었습니다. ABSTRACT Ring Mouse: A Wireless Input Device for Computers by Finger Motions By Phan Ke Hien Advisor: Professor Jung Doo Hee Course for Electronics Engineering Graduate School Korea Polytechnic University In this thesis, we studied a wireless device for the computer which is carried on the finger like a ring. Ring Mouse is a smart device that acts as a mouse that is connected wirelessly to a computer via Bluetooth. One can control the mouse cursor on the computer with finger motions. By using nRF52832 with microcontroller Cortex M4, we made a circuit that is power-efficient and cost-effective. Motion data is acquired by using an MPU-9250 module. The mechanical case of the device is designed and made by 3D printing technologies.

Chemical vapor deposition growth and high-k dielectric properties of hydrofluorocarbon films

PHAN THI KIM UYEN 忠南大學校 大學院 2024 국내석사

Chemical vapor deposition growth and high-k dielectric properties of hydrofluorocarbon films Phan Thi Kim Uyen Department of Materials Science and Engineering, Graduate School Chungnam National University Daejeon, Korea (Supervised by Professor Eui-Tae Kim) High-k dielectrics are crucial parts of the present generation and prospective electronic circuits. In complementary metal-oxide-semiconductor (CMOS) structures, particularly, they have been broadly studied for modern memory devices and logic microelectronics. In accordance with Moore's rule, an integrated circuit's transistor count doubles every two years, and the size of the transistors also shrinks predictably. To maintain transistor functionality, the gate dielectric thickness will be reduced to less than the oxide equivalent thickness (EOT) of a deca-nanometer, leaving high-k materials as the only practical option for such a small-scale EOT. However, for sub- 10nm thickness of about 1-2nm is beyond their elementary material limits forced by electron tunneling effect. Most of studies on high-k dielectric materials have been concentrated on inorganic thin film including metal oxides (MOs), nitrides (Si3N4, AlN), perovskites, and hybrids comprising them. Intriguingly, novel class of propitious candidates are organic such as fluorocarbon (FC), hydrocarbon (HC) or the most recent hydrofluorocarbon (HFC), irrespective of the fact how low-k they are commonly considered. An amorphous carbon monolayer, for example, is a strong insulator. HFC films (HFCs) have carbon backbones with hydrogen and fluorine atoms that distribute randomly along different sides of the backbone carbon atom. This affects the electric and dielectric properties of an HFC. As a result, HFCs with high k values show promise. A brief literature review on ultrathin high-k dielectric films of HFC films were fabricated and their exceptional features were utilized to develop the applications in metal-insulator-semiconductor (MIS) devices. For further application, they were employed as surface passivation layers in graphene field-effect transistors (GFET) to suppress Coulomb scattering from adsorbing ambient environment of graphene channel fabricated by inductively-coupled plasma chemical vapor deposition (ICP- CVD) using the plasma gas mixtures (CH4, CF4 and H2). This approach supplies a straightforward and commercial fabrication of high-k dielectric thin films for prospective applications in microelectronic devices. The surface passivation of HFC films for the enhanced mobility of graphene channel. This emergent structure is an auspicious candidate used as affordable, effective, and simple high-k dielectric films for alternatives of CMOS devices to provide insights for future research.

A Study on the volume estimation of household objects using deep learning

Phan Thai Trung Hongik University, Graduate School 2019 국내석사

Artificial Intelligence in general as well as Deep Learning in particular is one of the most tremendous techniques in the Fourth Industrial Revolution. By virtue of this, human can do what we have never thought that we can do in the past. Applications of Deep Learning are countless so far. One of applications the author of this thesis wants to emphasize here is to calculate the volume of household objects based on Deep Learning for Object Detection. As a result, we will have an objective and concise view about Deep Learning for Object Detection such as knowing about the latest advanced algorithms to solve the problems of Object Detection, the good frameworks or backends we should use for our algorithms in each special case, the newly architectures with much more accuracy and other related techniques as well. Furthermore, we also built a whole system from Android APP to Server running object detection algorithm and Database completely to simulate and visualize the theories we presented. Generally speaking, the author of this thesis presented one of the most accurate and reliable methods which can solve the problem of volume calculation of household objects based on Deep Learning for Object Detection. Hopefully, this can help and inspire others to build their own system to apply in real life.

Reciprocity approach for determination of surface wave motion and its application in scattering of surface waves by cavities

Phan, Haidang 부산대학교 2013 국내박사

Closed-form solutions of surface wave motions generated by a time-harmonic load applied in the interior of a half-space are determined in a simple manner by the use of reciprocity in elastodynamics. The method requires expressions for the displacements and the stresses of free surface waves, preferably in analytical form, but numerically obtained forms can also be used. A virtual wave that satisfies appropriate conditions on the boundaries and is a solution of the elastodynamic equations is used as state in the reciprocity theorem. By choosing a suitable virtual wave, state , which is the actual solution can be solved directly from the reciprocity relations. The solutions of surface wave motions are also obtained by the use of integral transform techniques. It is then shown that the amplitudes of the scattered waves obtained by the reciprocity approach and the integral transform approach are mathematically identical. Based on the obtained solutions of surface wave motions, a mathematical model for scattering of surface waves by a cavity at the surface of a half-space is investigated. The purpose of this study is therefore to introduce a novel theoretical approach for detection and characterization of cavities. The amplitudes of the scattered field are verified by the numerical results from the boundary element method (BEM). The analytical and BEM result are graphically displayed and show excellent agreement when the depth and the width of the cavity are small compared to the wavelength. Both results are then compared with the experiment data. As an improvement, a model for multiple scattering of surface waves by cavities on the surface of a half-space is studied. For multiple scattering by cavities, the self-consistent method is employed to derive an implicit set of equations which approximates the scattered field. Numerical calculations based on the boundary element method are found in order to compare with the analytical results. Comparisons between the analytical and BEM results provide good agreement.

Preparation of Near-Infrared-Absorbing Nanoparticles for Cancer Bioimaging and Photothermal Treatment : 암에 대한 바이오이미징 및 광열 치료법을 위한 근적외선 흡수 나노입자 제조

PHAN, THI TUONG VY 부경대학교 대학원 2019 국내박사

Cancer is among the leading causes of death worldwide. Cancer theragnosis agents with both cancer diagnosis and therapy abilities would be the next generation of cancer treatment. Recently, nanomaterials with strong absorption in near-infrared (NIR) region have been explored as promising cancer theragnosis agents for bio-imaging and photothermal therapy. In this dissertation, we reported the preparation of several advanced nanomaterials systems with biocompatible, strong NIR-absorbing properties. We offered the novel green method to synthesis NIR-absorbing agents to alter the costly and toxic traditional methods. The in vitro and in vivo photothermal anticancer activity results of the designed nanoparticles evidenced their promising potential in cancer treatment. The nanoparticles also gave the good amplitude of photoacoustic signals, which facilitates the imaging of tumor tissues using a non-invasive photoacoustic tomography system. Thus, our works highlight the great potential of using NIR-absorbing agents as theranostic nanoplatforms for cancer imaging-guided therapy.

Impact of mismatch lesion on long-term outcome of intravascular ultrasound guided percutaneous coronary intervention

Phan, Tan Quang 중앙대학교 대학원 2019 국내박사

연구목적 본 연구는 참조혈관 크기와 목표병변의 혈관크기가 다른 불일치병변 (mismatch lesion)이 혈관내 초음파를 이용한 경피적 관동맥 중재술의 결과에 미치는 영향을 평가하기위해 하였다. 연구방법 참조혈관 크기와 목표병변의 혈관크기가 다른 불일치병변은 근위부와 원위부 참조혈관 내강이 최소 1.0 mm 이상 또는 가장 작은 참조혈관 내강이 30 % 이상 차이가 나는 경우는 불일치혈관군(mismatch group)으로 하였고, 차이가 없었던 군은 일치혈관군 (non-mismatch group)으로 하였다. 총 1,706 병변을 후향적으로 분석하여 불일치혈관군 411 병변과 일치혈관군 1,295 병변으로 나누었다. 성향점수분석 (Propensity score matching) 후 최종 분석을 위해 각 군에서 397 병변을 선택하였고, 주요 심장 사건으로 사망, 심근 경색 및 목표병변 재개통술을 평균 5년간 관찰하여 분석하였다. 결과 대상군의 임상적 특성 및 혈관 조영 특성은 성향점수매칭후 두 군 간에 차이가 없었다. 일치혈관군과 비교했을 때, 불일치혈관군은 시술 시간이 더 길었고 (53.8±16.2 vs. 45.2±10.7 분, p<0.001), 조영제 사용량 (179.6±53.2 vs. 161.4±31.2 ml, p<0.001) 과 스텐트의 수(1.19±0.43 vs. 1.07±0.25, p<0.001) 가 더 많았다. 불일치혈관군은 스텐트시술후 추가적인 풍선확장술 최적화 치료가 더 많은 경우에서 필요하였고 (79.6% vs. 53.9%, p<0.001), 풍선확장시간이 길었으며 (2.98±1.75 vs. 1.91±0.92, p<0.001) 및 사용한 최대압력이 (19.5±3.9 vs. 16.7±3.7 atm, p<0.001) 더 높았다. 관동맥 중재술 동안, 불일치혈관군은 주요 합병증이 (7.8% vs. 4.3%, p=0.037) 더 높았고, 시술 성공률이 낮았다 (91.4% vs. 95.5%, p=0.022). 최종 혈관조영상에서, 불일치혈관군은 최소 내강 직경 (2.62±0.45 vs. 2.90±0.57 mm, p<0.001)이 작았고, 잔존 협착이 더 높았고 (11.3±4.5% vs. 9.4±6.2%, p<0.001), 시술 성공률이 낮았다 (93.2% vs. 96.7%, p=0.023). 최종 혈관내초음파에서 불일치혈관군은 최소 스텐트 면적이 더 작았고 (6.39±2.09 vs. 6.93±2.77 mm2, p=0.002), 불완전한 스텐트 밀착 및 스텐트 가장자리 박리가 (6.3% vs. 3.0%, p=0.029 and 2.5% vs. 0.8%, p=0.045)을 더 많았다. 하지만 스텐트 혈전증은 두 군 간에 유의한 차이가 없었다 (p=0.317). 추적 관찰 결과 불일치혈관군은 심장사망 (HR: 2.54, 95%CI:1.36-4.75, p=0.003)과 목표혈관 재개통술 (HR: 2.73, 95%CI:1.59-4.71, p<0.001)에 주로 기인한 주요심장사건(HR: 2.23, 95%CI:1.56-3.12, p<0.001)의 위험이 더 높았다. 결론 불일치 혈관병변은 관동맥 중재술이 가능한 병변이지만, 관동맥 중재술에 혈관내초음파를 사용하는데도 바람직하지 않은 시술결과와 나쁜 임상경과를 보였다. 본 연구 결과는 협착형태에서 불일치 혈관병변의 영향을 분석하였고, 적절한 접근법을 찾기 위해 전향적 연구에서 검증이 필요하다. Objectives To evaluate the impact of mismatch lesion on the outcome of percutaneous coronary intervention (PCI) with intravascular ultrasound (IVUS) guidance. Material and Methods Mismatch lesion was defined as the difference between the proximal and distal reference lumen diameters (RLD) of ≥1.0 mm or ≥30% of the smallest RLD. Total 1,706 lesions were retrospectively analyzed and divided into two groups, including 411 lesions in Mismatch group and 1,295 lesions in non-mismatch group. After propensity score matching, 397 lesions in each group were selected for final analysis. Major adverse cardiac events (MACE) was the composite of death, myocardial infarction and target lesion revascularization (TLR). Results The baseline and angiographic characteristics of participants were well balanced after propensity score matching. Compared with non-mismatch group, mismatch group PCI needed more procedure time (53.8±16.2 vs. 45.2±10.7 minutes, p<0.001), contrast amount (179.6±53.2 vs. 161.4±31.2 ml, p<0.001) and number of coronary stent (1.19±0.43 vs. 1.07±0.25, p<0.001). Mismatch group also required more frequently use of post-stenting optimization (79.6% vs. 53.9%, p<0.001) with more inflation times (2.98±1.75 vs. 1.91±0.92, p<0.001) using higher maximum pressure (19.5±3.9 vs. 16.7±3.7 atm, p<0.001). During PCI, mismatch group encountered more major complications (7.8% vs. 4.3%, p=0.037) and lower procedure success rate (91.4% vs. 95.5%, p=0.022). On final angiogram, mismatch group had smaller minimum lumen diameter (2.62±0.45 vs. 2.90±0.57 mm, p<0.001), higher residual stenosis (11.3±4.5% vs. 9.4±6.2%, p<0.001) and lower angiographic success rate (93.2% vs. 96.7%, p=0.023). On final IVUS, mismatch group achieved smaller minimal stent area (6.39±2.09 vs. 6.93±2.77 mm2, p=0.002) along with higher rate of incomplete stent apposition and stent edge dissection (6.3% vs. 3.0%, p=0.029 and 2.5% vs. 0.8%, p=0.045, respectively). Follow-up data during median of 5 years showed that mismatch group had higher risk of MACE (HR: 2.23, 95%CI:1.56-3.12, p<0.001), mainly driven from cardiac death (HR: 2.54, 95%CI:1.36-4.75, p=0.003) and TLR (HR: 2.73, 95%CI:1.59-4.71, p<0.001). Conclusions Mismatch lesion posed significant challenging during PCI that led to unfavorable procedural results and worse clinical outcomes even with IVUS guidance. These results need to be validated in prospective studies to determine the role of mismatch lesion in stenosis morphology classification and to find the suitable approach to deal with this lesion type.

Does a leader pay a price for being servant? the negative effects of servant leadership behaviors on leaders' well-being

Phan Thanh Thao Graduate School, Korea University 2021 국내석사

This study adopts job demands-resources theory as an overarching framework to examine the enactment of a servant leadership style and its potential negative effects on leaders themselves. Whereas much of the existing research literature about servant leadership focuses on how these leaders can improve follower and organizational outcomes, this study specifically investigates how this leadership style may act as an additional demand on the leaders who engage in it. Based on temporal separation data collected in Vietnam, the results indicated that engaging in servant leadership leads to emotional exhaustion by generating psychological job demand in terms of role overload. Moreover, the study identifies a boundary condition on the extent to which adopting servant leadership behaviors would be more stressful, by taking into account their followers’ characteristics. Specifically, the effect of servant leadership on leaders’ well-being is more strongly negative when leaders’ perception of subordinates’ manipulative intention is high. Theoretical, practical implications and directions for future research are discussed.