Mobile users can open program in new tab for better viewing.

Open program in new tab

Day 1 27/08/2020
Room #1

Organizing Committee Welcome 08:00 - 09:00

Starts at 9:00 Vietnam time

Welcome message by Conference Manager 08:10 - 09:00

Welcome message by EAI Community Manager 08:15 - 09:00

Keynote: Prof. Dong-Seong Kim 08:30 - 09:00

Reliability and Real-time Issues in Industrial IoT

Coffee Break 08:50 - 09:00

Session 1: Telecommunications Systems and Networks 09:00 - 10:00

09:00 - 09:00
Intelligent Channel Utilization Discovery in Drone to Drone Networks for Smart Cities

Dronenetworksareplayingasignificantroleinawidevariety of applications such as the delivery of goods, surveillance, search and res- cue missions, etc. The development of the drone to drone (D2D) networks can increase the success of these applications. One way of improving D2D network performance is the monitoring of the channel utilization of the link between drones. There are many works about monitoring channel utility; however, either they sense channel physically, which is not reli- able and effective due to noise in the channel and miss-sense of signals, or they have protocol-based solutions with high time-complexity. Hence, we propose a less time and power-consuming MAC layer protocol based monitoring model, which works on the IEEE 802.11 RTS/CTS protocol for D2D communication. We work on this protocol because it solves the hidden terminal problem, which can be seen widely in drone commu- nication due to the characteristics of wireless networks and mobility of drones. Our model consists of Searching & Finding and Functional Sub- layers. with our model, we decrease the sensing time of the channel by about 25%, and we reduce the power consumption of sensing drone approximately 26%. Also, our model uses about 57% less area during the calculation phase.
Authors: Muhammed Raşit Erol (Istanbul Technical University, Computer Engineering Dept.), Berk Canberk (Istanbul Technical University, Computer Engineering Dept.),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Downlink Resource Sharing and Multi-tier Caching Selection Maximized Multicast Video Delivery Capacity in 5G Ultra-dense Networks

In this paper, we propose a downlink resource sharing and multi-tier caching selection (DRS-MCS) solution for video streaming applications and services (VASs) in 5G ultra-dense networks (UDNs). The DRS-MCS allows mobile users (MUs) to experience the VASs by multicasting from three-tier caching placements, i.e., macro base station (MBS), femtocell base stations (FBSs), and mobile devices. To do so, the MUs are categorized into three types including 1) sharing users (SUs) that own downlink resources being shared for device-to-device (D2D) communications, 2) caching helpers (CHs) that cache the requested videos for multicasting over D2D communications, and 3) requesting users (RUs) that request the videos. The CHs and the RUs are grouped into different clusters, each cluster has a number of CHs and RUs in close vicinity for D2D multicast communications. We then formulate the DRS-MCS optimization problem. By solving the problem, the DRS-MCS solution can select not only the best pairs of SUs and CHs for D2D multicast communications but also the best caching placements for multicasting in each cluster, so as to maximize the total video capacity delivered to the RUs. Simulation results are shown to demonstrate the benefits of the proposed DRS-MCS solution compared to other conventional multicasting schemes.
Authors: Nguyen-Son Vo (Duy Tan University, Vietnam), Thanh-Minh Phan (Ho Chi Minh City University of Transport, Vietnam), Minh-Phung Bui (Van Lang University, Vietnam), Quang-Nhat Tran (Van Lang University, Vietnam), Hien Nguyen (Duy Tan University, Vietnam), Antonino Masaracchia (Queen's University Belfast, UK),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Performance Analysis of Relay Selection on Cooperative Uplink NOMA Network with Wireless Power Transfer

Wireless power transmission in the next-generation wireless networks is the subject that attracts a lot of attention from academia and industry. In this work, we study and analyze the performance of relay selection on uplink non-orthogonal multiple access (NOMA) networks with wireless power transmission. Specifically, the considered system consists of one base station, multiple power-constrained relays and a pair of NOMA users. The best relay (with highest energy harvested from the base station) is chosen to cooperate with two users which use NOMA scheme to send messages to the base station. To analyze the performance, based on the statistical characteristics of signal-to-noise ratio (SNR) and signal-to-interference-plus-noise ratio (SINR), using the Gaussian-Chebyshev quadrature method, the closed-form expressions of outage probability and throughput for two users are derived. In order to understand more details about the behavior of this considered system, the numerical results on outage probability and throughput of a given system are provided following the system key parameters, such as the transmit power, the number of relays, time switching ratio and energy conversion efficiency. In the end, the theoretical result is also verified by using the Monte-Carlo simulation. The simulation results demonstrate that the performance of the system is improved by increasing the number of relays
Authors: Long Nguyen Van (dtu.edu.vn), Truong Truong (Duy Tan University), Dac-Binh Ha (Duy Tan University), Loc Vo (Pham Van Dong University), Lee Yoonill (Purdue University Northwest),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Convolutional Neural Network-based DOA estimation Using Non-uniform Linear Array for Multipath Channels

In this paper, a novel convolutional neural network (CNN) design was proposed for DOA estimation, which could deploy in radio-electronics systems for im-proving the accuracy and operation efficiency. The proposed model was evaluated with different hyper-parameter configurations for optimization, and then a suitable model was compared with other existing models to demonstrate its preeminence. Regarding dataset generation, our work considered the influence of both Gaussian noise and multipath channels to DOA estimation accuracy. Moreover, a non-uniform linear array model was proposed to outperform a uniform linear array with the same number of elements and operation conditions when ap-plying the same CNN model. According to the analysis, the model with 5 conv-blocks, 48 filters, and a filter size of 1×7 achieved the best performance in terms of accuracy (75.27% at +5dB SNR) and execution time (10.1 ms) that notably outperformed two other state-of-the-art CNN model-based DOA estimation techniques.
Authors: Van Sang Doan (Kumoh National Institute of Technology), Thien Huynh-The (Kumoh National Institute of Technology), Van-Phuc Hoang (Le Quy Don Technical University), Dong Seong Kim (Kumoh National Institute of Technology),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
An UAV and Distributed STBC for Wireless Relay Networks in Search and Rescue Operations

This paper proposes a transmission method using unmanned aerial vehicle (UAV) with distributed Space-Time Block Code (STBC) for multi-hop wireless relay networks in search and rescue operations. First, an UAV is considered to add to the hop with the minimum output signal-noise-ratio (SNR) and operates as a relay node to maintain the links between adjacency nodes in network, expand the transmission coverage area and improve the transmission performance. In addition, in order to overcome the difficulty in assigning the STBC patterns to the distributed relays and also alleviate the complexity of system design and implementation, the original STBC pattern is modified while keeping the same cooperative diversity gain. Finally, an algorithm is proposed to find out the optimal location of the added UAV in the hop, where the UAV has the best contribution to the data transmission performance between the transmitter and the receiver. It can be seen from the simulation results, the optimal location of the added UAV depends on not only the environment of real scenarios but also the distributed cooperative diversity gain. We can confirm that the proposed method achieves the significant performance improvement while keeping the simple operation of system for UAV communications in search and rescue operations.
Authors: Cong Hoang Diem (Department of Computer Networks, Faculty of Information Technology, Hanoi University of Mining and Geology (HUMG)), Takeo Fujii (The University of Electro-Communications),
Hide Authors & Abstract

Show Authors & Abstract

Lunch 09:00 - 09:00

30 minutes

Session 2: Hardware, Software, and Application Designs 09:00 - 09:00

09:00 - 09:00
Resolution-improvement of confocal fluorescence micros-copy via two different point spread functions

In this paper, we propose a new method to obtain the improvement of lateral axial resolution of confocal fluorescence microscopy. In this method, we employ two dif-ferent beams to illuminate the sample: (1) the Gaussian beam; (2) the donut beam. Two different images are produced from these two illumination beams. A higher res-olution image is generated by a multi-relationship between these two image. A set of simulation and experimental results are employed to compare the effectiveness of proposed method with the traditional confocal fluorescence microscopy. These re-sults demonstrated that our method can be employed to achieve the resolution-enhancement of confocal fluorescence microscopy.
Authors: MinhNghia Pham, XuanHoi Hoang (Le Quy Don Technical University), VanNhu Le (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Estimations of Matching Layers Effects on Lens Antenna Characteristics

The dielectric lens antenna is a prime candidate for the mm-wave 5G communica-tions system. The size and the radiation efficiency can be improved by using a high-density dielectric lens antenna. However, the dense dielectric material lenses can make some antenna properties deteriorate due to the reflections at the surface between the air and the dielectric. These unexpected effects can be solved by us-ing a quarter-wavelength matching layer (ML). In this paper, the authors perform the study to estimate the influence of the ML on the antenna properties on specific dielectric materials. The results illustrate a marked improvement in gain, a signifi-cant reduction in the side-lobe level, and considerable changes in the electric field distribution on the plane with and without using the ML. Besides, the article also shows the abilities to minimize the antenna size when different types of dielectric materials are chosen while maintaining the antenna radiation characteristics.
Authors: Hung Phan (Le Quy Don Technical University), Nguyen Quoc Dinh (Le Quy Don Technical University), Thuyen Hoang (Le Quy Don Technical University), Hung Nguyen (Le Quy Don Technical University), Thuy Le (Hanoi University of Science and Technology), Trung Le (Telecommunications University), Yoshihide Yamada (Malaysia-Japan International Institute of Technology UTM),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
A 3-stacked GaN HEMT Power Amplifier with Independently Biased Technique

In this paper, a design of 3-stacked GaN high-electron-mobility transistor radio-frequency power amplifier using independently biased technique is presented. The power amplifier operates at 1.6 GHz for wireless communications applications. The possibility of independently adjusting operation conditions for each transistor of the proposed amplifier brings a promising advantage of efficiency improvement. By independently setting proper bias conditions, DC power consumption of the power amplifier can be reduced leading to efficiency enhancement without output power degradation. A performance comparison of the proposed power amplifier with a conventional 3-stacked power amplifier has been performed. The simulated results indicate that the proposed power amplifier offers superior efficiency over the conventional one.
Authors: Duy Manh Luong (Le Quy Don Technical University), Huong Tran (Le Quy Don Technical University), Doanh Bui (Le Quy Don Technical University), Son Vo (University of Communications and Transport),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Feasibility and Design Trade-offs of Neural Network Accelerators Implemented on Reconfigurable Hardware

In recent years, neural networks based algorithms have been widely applied in computer vision applications. FPGA technology emerges as a promising choice for hardware acceleration owing to high-performance and flexibility; energy-efficiency compared to CPU and GPU; fast development round. FPGA recently has gradually become a viable alternative to the GPU/CPU platform. This work conducts a study on the practical implementation of neural network accelerators based-on reconfigurable hardware (FPGA). This systematically analyzes utilization-accuracy-performance trade-offs in the hardware implementations of neural networks using FPGAs and discusses the feasibility of applying those designs in reality. We have developed a highly generic architecture for implementing a single neural network layer, which eventually permits further construct arbitrary networks. As a case study, we implemented a neural network accelerator on FPGA for MNIST and CIFAR-10 dataset. The major results indicate that the hardware design outperforms by at least 1500 times when the parallel coefficient p is 1 and maybe faster up to 20,000 times when that is 16 compared to the implementation on the software while the accuracy degradations in all cases are negligible, i.e., about 0.1% lower. Regarding resource utilization, modern FPGA undoubtedly can accommodate those designs, e.g., 2-layer design with p equals 4 for MNIST and CIFAR occupied 26% and 32% of LUT on Kintex-7 XC7K325T respectively.
Authors: Kien Trinh (Le Quy Don Technical University), Manh Duong (Le Quy Don Technical University), Nga Dao (Le Quy Don Technical University), Thanh Nguyen (Posts and Telecommunications Institute of Technology), Phong Nguyen (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract

Coffee Break 09:00 - 09:00

10 minutes

Session 3: Information Processing and Data Analysis 09:00 - 09:00

09:00 - 09:00
Adaptive Essential Matrix based Stereo Visual Odometry

Visual Odometry is widely used for recovering the trajectory of a vehicle in an autonomous navigation system. In this paper, we present an adaptive stereo visual odometry that separately estimates the rotation and translation. The basic framework of VISO2 is used here for feature extraction and matching due to its feature repeatability and real-time speed on standard CPU. The rotation is accurately obtained from the essential matrix of every two consecutive frames in order to avoid the affection of the stereo calibration uncertainty. With the estimated rotation, translation is rapidly calculated and refined by our proposed linear system with non-iterative refinement without the requirement of any ground truth data. The further improvement of the translation by joint backward and forward estimation is also presented in the same framework of the proposed linear system. The experimental results evaluated on the KITTI dataset demonstrate around 30% accuracy enhancement of the proposed scheme over the traditional visual odometry pipeline without much increase in the system overload
Authors: Huu Hung Nguyen (Le Quy Don Technical University), Quang Thi Nguyen (Le Quy Don Technical University), Manh Tran (Le Quy Don Technical University), Dong-Seong Kim (Kumoh National Institute of Technology),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
A modified localization technique for pinpointing a gunshot event using acoustic signals

This paper proposes a method for localizing a gunshot event using four acoustic sensor nodes mounted at the four corners of a rectangular working area. Each of these nodes involves three sensors to acquire acoustic signals of any gunshot inside the working area. The approach analyzes individual signals received by the nodes to identify sound events using false alarm probability and determine their emission directions exploiting a minimum mean square error estimator and the time difference of arrival of the events. The gunshot location is the quadrilateral center of four crossing points resulting from pairs of adjacent event emission directions. For evaluating the proposed method, a signal including ten real gunshots recorded by a nearby acoustic sensor is delayed and attenuated according to a theoretical wave propagation model to create various signal patterns, which simulates signals received by the installed sensor nodes. Furthermore, the Gaussian noise is added to the simulated signals to emulate the influence of wave propagation environment. This article also implements some mechanisms to compute the time difference of arrival for comparison. They are comprised of the first crossing of threshold and signal, maximum amplitude, Akaike’s Information Criterion, and the cross correlation function. Hence, one of them can be selected for a real application. Experimental results show that the proposed method achieves high accuracy of gunshot localization.
Authors: Thin Tran (Military Institute of Science and Technology), My Bui (Military Institute of Science and Technology), Hoang Nguyen (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Table Structure Recognition in Scanned Images using a Clustering Method

OCR for scanned paper invoices is very challenging due to the variability of 19 invoice layouts, different information fields, large data tables, and low scanning quality. In this case, table structure recognition is a critical task in which all rows, columns, and cells must be accurately positioned and extracted. Existing methods such as DeepDeSRT only dealt with high-quality born-digital images (e.g., PDF) with low noise and apparent table structure. This paper proposes a novel and efficient method named CluSTi. The contributions of CluSTi are three-fold. Firstly, it removes heavy noises in the table images using a clustering algorithm. Secondly, it extracts all text boxes using state-of-the-art text recognition. Thirdly, based on the horizontal and vertical clustering algorithm with optimized parameters, CluSTi groups the text boxes into their correct rows and columns, respectively. The method was evaluated on three datasets: i) 397 public scanned images; ii) 193 PDF document images from ICDAR 2013 competition dataset; and iii) 281 PDF document images from ICDAR 2019's numeric tables. The evaluation results showed that CluSTi achieved an F1-score of 87.5%, 98.5%, and 94.5%, respectively. Our method also outperformed DeepDeSRT with an F1-score of 91.44% on only 34 images from the ICDAR 2013 competition dataset. To the best of our knowledge, CluSTi is the first method to tackle the table structure recognition problem on scanned images.
Authors: Nam Nguyen Van (Thuyloi University), Hanh Vu (Viettel CyberSpace Center), Arthur Zucker (Sorbonne University, Polytech Sorbonne), Younes Belkada (Sorbonne University, Polytech Sorbonne), Hai Do Van (Thuyloi Univeristy), Nam Nguyen Van (Thuyloi University), Doanh Nguyen-Ngoc (Thuyloi University), Thanh Le Nguyen Tuan (Thuyloi Univerisy), Dong Hoang Van (Thuyloi University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Distributed Watermarking for Cross-domain of Semantic Large Image Database

This paper proposes a new method of distributed watermarking for large image database that is used for deep learning. We detect the semantic meaning of set of images from the database and embed the a part of watermark into such image set. Since the image sets have multiple image and are distributed in the whole of multiple database, we expect that the proposed method is robust against several attacks.
Authors: Ta Minh Thanh (Le Quy Don Technical University), Le Danh Tai (Le Quy Don Technical University), Nguyen Kim Thang (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Depth Image Reconstruction using Low Rank and Total Variation Representations

Rapid advancement and active research in computer vision applications and 3D imaging have made a high demand for efficient depth image estimation techniques. The depth image acquisition, however, is typically challenged due to poor hardware performance and high computation cost. To tackle such limitations, this paper proposes an efficient approach for depth image reconstruction using low rank (LR) and total variation (TV) regularizations. The key idea is LR incorporates non-local depth information and TV takes into account the local spatial consistency. The proposed model reformulates the task of depth image estimate as a joint LR-TV regularized minimization problem, in which LR is used to approximate the low-dimensional structure of the depth image, and TV is employed to promote the depth sparsity in the gradient domain. Furthermore, this paper introduces an algorithm based on alternating direction method of multipliers (ADMM) for solving the minimization problem, whose solution provides an estimate of the depth map from incomplete pixels. Experimental results are conducted and the results show that the proposed approach is very effective at estimating high-quality depth images and is robust to different types of data missing models.
Authors: Van Ha Tang (Le Quy Don Technical University), Mau Uyen Nguyen (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Deep Learning Based Hyperspectral Images Analysis for Shrimp Contaminated Detection

In this paper, a deep learning based hyperspectral image analysis for detecting contaminated shrimp is proposed. The ability of distinguishing shrimps into two classes: clean and contaminated shrimps is visualized by t-distributed Stochastic Neighbor Embedding (t-SNE) using spectral feature data. Using only some small data set of hyperspectral images of shrimps, a simple processing technique is applied to generate enough data for training a deep neural network (DNN) with high reliability. Our results attain the accuracy of 98% and F1-score over 94%. This works confirms that with only few data samples, Hyperspectral Imaging processing technique together with DNN can be used to classify abnormality in agricultural productions like shrimp.
Authors: Minh Hieu Nguyen, Huyen Nguyen (Hanoi University of Science and Technology), Cong Nguyen Pham (Hanoi University of Science and Technology), Ngoc Lê (Hanoi University of Science and Technology), Huy Dung Han (Hanoi University of Science and Technology),
Hide Authors & Abstract

Show Authors & Abstract
Day 2 28/08/2020
Room #1

Session 4: Industrial Networks and Intelligent Systems 09:00 - 09:00

09:00 - 09:00
An optimal eigenvalue-based decomposition approach for forest parameters retrieval over hilly forest areas

The major purpose of this paper is providing a new method for retrieving forest parameters in mountain forest areas using L-band polarimetric synthetic aperture radar interferometry (PolInSAR) data. The new way for forest parameters extrac-tion is the application of the target decomposition technique to PolInSAR data. However, the modeling of the vegetation backscattering mechanisms is compli-cated due to the influences of the topographic slope variation and assumptions about the volume scattering component. In order to overcome these drawbacks, an eigenvalue-based decomposition technique with the simplified Neumann vol-ume scattering model is suggested for forest parameters retrieval. The proposed method is applied to the both simulated data and ALOS/PALSAR spaceborne da-ta for evaluating the effectiveness of proposed manner. Experimental results indi-cate that the proposed method provides a more accurate and reliable results than 3-stage method in estimating forest parameters on steep terrain.
Authors: MinhNghia Pham (Le Quy Don Technical University), Nguyen Tan (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
An improved forest height inversion method using dual-polarization PolInSAR data

The Extended 3-stage method has somewhat improved the precision of estimating forest parameters of the traditional 3-stage inversion method. However, the to-pography phase and optimal volume coherence coefficient determined by this method are not really optimal, that lead to the forest parameters estimation of this method is unstable and inaccuracy. Therefore, this paper proposes an improved forest height inversion method by dual-polarization channel PolInSAR data to enhance the precision of forest parameters extraction. In the suggested approach, the surface phase is calculated based on the mean coherence set theory. A com-prehensive search method is then proposed to determine the polarization channels corresponding to the optimal polarization channel coherence coefficients for the volume scattering component. The effectiveness of the suggested approach was assessed with simulation data from PolSARprosim 5.2 software. The empirical results show that the suggested approach not only improves the effectiveness of the forest parameters estimation of the extended 3-stage method but also reduces the complexity in the calculation.
Authors: MinhNghia Pham (Le Quy Don Technical University), Cuong Huu (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
An Attempt to Perform TCP ACK-Storm on Virtual and Docker Network

Recently, the server virtualization (hypervisor) market is growing up fast because server virtualization has many benefits. More and more businesses use hypervisors as an alternative solution to a physical server. However, hypervisors are more vulnerable than traditional servers according to recent researches. Therefore, stand on the position of a system administrator, it's necessary to prepare for the worst circumstances, understand clearly, and research for new threats that can break down the virtual system. In this paper, we attempt to perform TCP ACK storm based DoS (Denial of Service) attack on virtual and docker network and propose some solutions to prevent them.
Authors: Ta Minh Thanh (Le Quy Don Technical University), Khanh Tran Nam (Le Quy Don Technical University), Thanh Nguyen Kim (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Identification of Chicken Diseases using VGGNet and ResNet Models

Nowadays, food security is essential in human life, especially for poultry meat. Therefore, the poultry raising is growing over years. This leads to the development of diseases on poultry, resulting in potentially great harm to human and the surrounding environment. It is estimated that when the diseases spread, the economic and environmental damages are relatively large. In addition, small-scale animal husbandry and an automated process to identify diseased chickens are essential. Therefore, this work presents an application of machine learning algorithms for automatic poultry disease identification. Here, the deep convolutional neural networks (CNNs) namely VGGNet and ResNet are used. The algorithms can identify four common diseases in chickens namely Avian Pox, Infectious Laryngotracheitis, Newscalte, and Marek against healthy ones. The obtained experimental results indicate that the highest achievable accuracies are 74.1% and 66.91% for VGGNet-16 and ResNet-50 respectively... The initial results showed positive results, serving the needs of the building and improving the model to achieve higher results.
Authors: Da Quach Luyl, Duc Chung Tran (Co-Authors), Nghi Pham-Quoc (Co-Authors), Mohd Fadzil Hassan (Co-Authors),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Design and evaluation of the grid-connected solar power system at the stage of DC BUS with optimization of modulation frequency for performance improvement

In grid-connected solar panel systems, the power converters play a very important role in control systems, because the characteristics of solar panel system are that the generation power is constantly changing due to dependence on weather conditions. This article presents the research results of the application of power electronic converter in grid-connected solar power system. In particular, we focus on building an algorithm to control and simulate the grid-connected solar power system at the DC-Bus stage by setting an optimal set of SVPWM (Space Vector Pulse Width Modulation) modulation frequencies when the pulse width values are different. The simulation results on Matlab simulink show that the system operates stably, ensures the requirements of DC-Bus grid integration. The set-up system operates safely, has a simple control structure and algorithm, is easy to calibrate and makes it ready for practical applications.
Authors: Duc Minh Nguyen, Duc-Cuong Quach (Hanoi University of Industry, Viet Nam), Ninh Nguyen (Institute of Energy Science - Viet Nam Academy of Science and Technology), Y Do (Hanoi University of Mining and Geology, Viet Nam), Chuong Trinh (Hanoi University of Industry, Viet Nam),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
A Predictive System for IoTs Reconfiguration Based on TensorFlow Framework

IoTs are rapidly growing with the addition of new sensors and devices to existing IoTs. Due to the ever-increasing demand for IoT nodes to adapt to changing environment conditions and application requirements, the need for reconfiguring these already existing IoTs is rapidly increasing. It is also important to manage the intelligent context to execute when it will trigger the appropriate behavior. Yet, many algorithms based on different models for time-series sensor data prediction can be used for this purpose. However, each algorithm has its own advantages and disadvantages, resulting in different reconfiguration behavior predictions for each specific IoTs application. Developing an IoTs reconfiguration application has difficulty implementing many different data prediction algorithms for different sensor measurements to find the most suitable algorithm. In this paper, we propose IoTs Reconfiguration Prediction System (IRPS), a tool that helps IoT developers to choose the most suitable time-series sensor data prediction algorithms for trigger IoT reconfiguration actions.
Authors: Tuan NGUYEN-ANH (University of Information Technology - Vietnam National University at HCM City), Quan LE-TRUNG (University of Information Technology - Vietnam National University at HCM City),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Building a Dynamic Predicted-Traffic-Congested-Discovery System to Resolve Chaotic Traffic Situations

Chaotic traffic situations usually link to traffic congestion. It poses many risks to commuters like traffic accidents, especially during bad weather. Therefore, building a system to anticipate congestion could enhance public safety and give traffic police forces time to handle traffic flows in potentially dangerous areas. Moreover, if types of predicted congestion can be formed in patterns, we can build reaction plans with various alert codes. They create dynamic risk maps that can provide useful knowledge to both authorities and travelers to make rescue and travel plans effectively. We introduce a dynamic predicted-congestion-pattern discovery system that integrates Fusion-3DCNN (a deep learning model) and PFP (periodic-frequent pattern) algorithm to predict congestion areas and discover congested-patterns. First, Fusion-3DCNN receives raster images whose channels contain a different type of data (e.g., rainfall, congestion, accidents) to predict N future periods (e.g., $T$, $T+1$, $T+N$) congestion. Then, for each period ($T+i$), the PFP is utilized from ($T+i$) back to ($T-j$) periods to discover the predicted-congestion patterns. The evolutions of congestion are visualized dynamically and predictably when arranging $T$ to ($T+N$) patterns on the road map. The experiments are performed using a dataset containing three sources of urban sensing data collected in Kobe City, Japan, from 2014-2015. From the experimental results, promising applications are discovered.
Authors: Ngoc-Thanh Nguyen (VNUHCM, University of Information Technology), Minh-Son Dao (National Institute of Information and Communications Technology), Koji Zettsu (National Institute of Information and Communications Technology),
Hide Authors & Abstract

Show Authors & Abstract

Coffee Break 09:00 - 09:00

10 minutes

Session 5: Security and Privacy 09:00 - 09:00

09:00 - 09:00
An Efficient Side Channel Attack Technique with Improved Correlation Power Analysis

Correlation Power Analysis (CPA) is an efficient way to recover the secret key of the target device. CPA technique exploits the linear relationship between the power model and the real power consumption of an encryption device. In fact, we only need fewer power traces to recover secret key bytes successfully. However, due to impact of noise, we need a larger number of power traces in order to extract the secret key. Therefore, the computation time becomes a serious problem for performing this attack. This paper introduces a new method to reduce timing computation for CPA method with the technique of finding points of interest which was used for template attack. The experimental results have clarified the efficiency of the proposed method.
Authors: Ngoc Tuan Do (Le Quy Don Technical University), Van-Phuc Hoang (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
An Optimal Packet Assignment Algorithm for Multi-level Network Intrusion Detection Systems

With the outbreaks of recent cyber-attacks, a network intrusion detection system (NIDS) which can detect and classify abnormal traffic data has drawn a lot of attention. Although detection time and accuracy are important factors, there is no work considering both contrastive objectives in an NIDS. In order to quickly and accurately respond to network threats, intrusion detection algorithms should be implemented on both fog and cloud devices, which have different levels of computing capacity and detection time, in a collaborative manner. Therefore, this work proposes a packet assignment algorithm that assigns detection and classification tasks for appropriate processing devices. Specifically, we formulate a novel optimization problem that minimizes detection time while achieving accuracy performance and computational constraints. Then, an optimal packet assignment algorithm that allocates as many packets as possible to fog devices in order to shorten the detection time is proposed. The experimental results on a state-of-the-art network dataset (UNSW-NB15) show that the proposed packet assignment algorithm produces similar performance to the optimal solution with regard to the detection time and accuracy.
Authors: Thi-Nga Dao (Le Quy Don Technical University), Duc Le (Nanyang Technological University), Hieu Ta (Le Quy Don Technical University), Son Vu (Le Quy Don Technical University),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
Privacy-Preserving for Web Hosting

Privacy-preserving for data is one of the crucial responsibilities of service pro-viders. For the web hosting service, customers’ information and source codes of websites are considered sensitive data that need to protect. The solution decentral-ized for web hosting is a new technology trend, provides an effective mechanism for storing and accessing websites. This paper addresses some problems related to privacy concerns of the decentralized web hosting service. Based on the block-chain technology, cryptography, and interplanetary file system (IPFS) platform, we propose a protocol for web hosting that provides three features. The first one ensures anonymity for information of customers. The second feature provides confidentiality and authentication for transmitting source codes of websites be-tween customers and the service provider (SP). And the last one is responsible for securing the source code of websites from other nodes on the public IPFS network. The experiments demonstrate that the proposed solution efficiently pro-tects privacy for the decentralized web hosting service.
Authors: Tam Huynh (Posts & Telecommunications Institute, HCMC, Vietnam), Thuc Nguyen (University of Science, HCMC, Vietnam), Nhung Nguyen (University of Science, HCMC, Vietnam), Hanh Tan (Posts & Telecommunications Institute, HCMC, Vietnam),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 09:00
A Novel Secure Protocol for Mobile Edge Computing Network applied Downlink NOMA

In this paper, we study a mobile edge computing (MEC) network based on non-orthogonal multiple access (NOMA) scheme, in which a user can offload its tasks to two MEC servers through downlink NOMA. Due to security constraints, the confidential tasks must be computed on the trusted server and the remain tasks can be offloaded to another server if needed. In this scenario, we propose a novel secure protocol, namely APS-NOMA MEC, based on access point selection (APS) scheme to guarantee the security constraint. The exact closed-form expression of successful computation probability for this proposed system protocol is derived. We further study the impact of the network parameters on the system performance to confirm the effectiveness of deployment of NOMA in MEC network. The numerical results show that our proposed protocol outperforms the conventional NOMA MEC scheme in terms of computation efficiency. Finally, the simulation results are also provided to verify the accuracy of our analysis.
Authors: Dac-Binh Ha (Duy Tan University), Truong Truong (Duy Tan University), Duy-Hung Ha (VSB-Technical University of Ostrava),
Hide Authors & Abstract

Show Authors & Abstract

Best Paper announcement 09:00 - 09:00