Earticle

Travel trends are changing due to the prolonged COVID-19 pandemic and vaccine development. The analysis of pre and post Covid-19 tourism trends, according to a survey by Jeju Tourism Organization, shows that the search volume for overseas travel has decreased compared to 2018 and 2019, but search volume for Jeju travel and the number of tourists visiting Jeju Island Increased. In the case of tourists in Jeju, many use vehicles, mostly rental cars, for transportation due to the geographical characteristics of the island, and the number of electric vehicles is increasing in Jeju Island’s rental car services due to the strengthening of electric car policies. However, most of the existing research on tourists have been conducted using public data. Therefore, based on the means of transportation mainly used by tourists, electric vehicle driving data recorded for three years provided by Korea Electric Power Corporation Knowledge Data Network (KEPCO KDN) was classified into a total of 11 areas by weather and time requirements and classified through an artificial intelligence-based multiclassification model. In this study, tourist activity patterns were classified according to season, time zone, and climate conditions, but in the future, it can be used for recommendations and advertisements for tourist destinations by subdividing zones and adding information on users.

The ciphertexts obtained by traditional encryption techniques is not totally random sequence forms. Many cryptanalytic studies based on mathematical analysis such as linear cryptanalysis and differential cryptanalysis have been conducted. Recently, deep learning-based cryptanalysis have been proposed to show more powerful attacks than the other mathematical-based approaches. In this paper, we propose a new automated deep learning-based approach to break encryption algorithms with Feistel structure.

To manage the big data in constraint resources has difficulties and challenges. The power and the cost can be saved when the cloud services are used to process and store the data. However, the data includes the personal information that can be sensitive and should be hidden from the others. So we propose the privacy-preserving classification scheme for image data. The pixel-based learning is the scheme that is adapted to the cryptosystem, and is used to classify the encrypted images. Our proposed deep learning model has the convolutional layers that has the same size of the kernel with the block size in the cryptographic algorithm. The experiment results show that it can improve the accuracy on classification of encrypted images, and make it possible to use the private data securely.

Although there have been studies using various algorithms on the delivery time prediction in the logistics business, those studies did not reflect various features such as region or product. In the case of delivery time prediction of a single model that does not reflect the features, the accuracy of delivery time prediction for a region with a high distribution is high, but the prediction accuracy is low for a region with a low distribution. To solve this problem, this paper proposes a method of classifying logistic patterns using clustering and selecting an optimal model for each logistic pattern. The proposed method consists of four steps. First, the derived variables such as reception day, delivery speed and delivery distance are created. Second, the data with the same pattern goes through clustering using K-means. Third, by comparing the performance of each model using six regression algorithms for each classified logistic pattern, an optimal model is selected and the model is stored. Lastly, the logistic pattern of the data to be predicted is classified and the optimal model stored for each pattern is loaded, and the result of delivery time prediction is provided through the model. Two experiments were performed to verify the proposed method. The e-commerce data from Brazil is used as verification data. From the experiment, the delivery time prediction error of the proposed model was smaller than that of the single regression model.

In order to solve the safety hazards caused by the operation failure of elevators, pump rooms, fire control facilities, access control and other equipment and facilities in colleges, this paper uses the Internet of Things technology to realize the ubiquitous access of equipment and facilities in colleges, and collect the operating status data of the equipment and facilities in real time. Establish an equipment and facility operation data center, use big data and deep learning technology to realize online monitoring and real-time early warning of equipment and facilities, and send the monitoring and early warning information to managers and equipment maintenance personnel for processing, and realize the monitoring, early warning, and processing of equipment and facilities Automated process management to build an intelligent monitoring system for college equipment and facilities.

In regard to recent advancements in metagenomic sequencing, it is now possible to sequence large numbers of microbial genomes with ease. Taxonomic classification of metagenomic data remains a crucial task as it provides useful information in finding relationships between other microbial species in a given area or possible infectious diseases. Over the past decade, deep learning has proven to be a powerful tool in classifying multiple objects. By combining both studies it is possible to gain taxonomic classification of metagenomic data with proficiency.

All over the world work-related musculoskeletal disorders are starting to take hold among digital device users, especially computer users. Weakness, sleepiness, shoulder pain, back pain/solidness/pain in the neck, and cerebral pain are quite often recorded as clear indicators during or after the use of personal computers, laptops, and devices. The most painful fact is that all these delicate indicators are routinely ignored as a sign of general weakness in most cases. This research work highlights this issue under the belt of Ergonomics. How can construct personal computers, laptops, and their associated equipment to be more human-friendly so that these serious health risks can be avoided, particularly in children. The paper also highlights previous important studies conducted at elementary levels of this subject.

Latest development, expansion, and organization in the field of Internet of things (IoT) are revolutionizing the day-byday life of individuals. To fulfill the extraordinary client needs, the 5G Networks are developing, and are significant for ceaseless development of IoT networks. The unfurling cellular innovation has brought the energy utilization up in mobile Networks with the carbon impression flooding to disturbing rates. This is causing an unfriendly impact on the climate and human wellbeing. The other challenge is lessening the general network dormancy and expanding throughput without forfeiting unwavering quality. One achievable option is conjunction of networks working on various frequencies. Notwithstanding, information data transmission backing, and range accessibility are the significant difficulties. Consequently, cognitive radio networks (CRN) are the possible solution to oblige every one of these difficulties for the concurrence deployment of 5G green cognitive radio networks for IoTs.

To provide the effective monitoring in health care field, the access of communication medium to transmitting nodes should be according to the medical conditions of patients. For more reliable and efficient use of WBANs the medium access needs to be assigned to data transmissions before the occurrence of emergency conditions, so that timely treatments can be triggered for medically critical patients. The existing MAC schemes are able to prioritize the emergency data for transmissions but, after the occurrence of emergency events and the prioritization process of data transmissions itself incurs delay. The delays in transmissions of medically critical patient’s data can be disastrous. Even the conventional MAC schemes of IEEE 802.15.4 and 802.15.6 protocols are not able to transmit the medical data with dynamic and predicted priorities. This research study focused on the prioritization of the data transmissions of the patients based on the predicted conditions of patients in dynamic manner. Therefore, the data of patients is analyzed to predict the emergency events and then prioritized to transmit the data of critical patients to medical concerns in time. The design of proposed MAC scheme is based on Time Division Multiple Access which makes it a congestion free MAC scheme. The proposed scheme is tested in simulation environment. The results showed that the presented MAC scheme predicted the patient’s condition with approximately 87% of accuracy and prioritized the data streams accordingly in dynamic manners. The proposed scheme efficiently prioritized the patient’s data transmissions when tested on one type of dataset and some factors are assumed ideal or null too.

The performance overhead is a major threat to cloud computing services. Of the many ways to reduce overhead, refreshing the oldest data in a cache is decent. In this paper, we designed hardware that allows cache clean on Arm architecture. We demonstrate 32 times better performance by leveraging the hardware CIH.

The processor's Hardware Performance Counter (HPC) is a special purpose register that counts hardware-related events. Although HPC monitoring is recognized to have less overhead than other profiling techniques, but it is still an important issue in using HPC with a short monitoring interval. In this paper, we profile and analyze the overhead of monitoring HPC according to measurement schemes: the number of events in groups, the number of simultaneously active monitoring targets, and the reading method. For the number of events in the group, overhead per event increases by 5-6%. In the case of the number of simultaneous active subjects, overhead increases by 0.8-1% per core and 0.1-0.2% per process. The read method has no consistent relationship with overhead. Therefore, in order to optimize HPC monitoring in consideration of overhead, after considering the number of events and the number of active monitoring targets in order, the reading method is selected according to the situation considering usability.

Since the VR HMD completely blocks the wearer's field of view, accidents can be prevented in advance by removing surrounding risk factors. In general, when using VR HMD content, a method of deploying safety personnel or securing an experience space is used. However, when an individual uses content, there is a problem that the method cannot be used. In this paper, the OpenCV library and the YOLO v4 algorithm were used to improve the above problems. An object may be detected from an input real-time image using the YOLO algorithm. Object detection was implemented using the front camera of the HTC VIVE Pro Eye equipment. Through the implemented technology, VR experiencers whose eyes were blocked were able to identify risk factors through the front camera.

To adapt to more Internet of Things (IoT) service scenarios, a new synchronous communication scheme was proposed in BLE 5.0. This solution enables devices to communicate through the periodic advertising synchronization establishment procedure (PASEP) without establishing a connection. PASEP can be divided into two steps. In the first step, a BLE device receives periodic advertising synchronization information from another device; in the second step, the BLE device periodically receives data using the received synchronization information. However, there are some problems in this process. In PASEP, there are only three channels available for use in the first step of PASEP, which leads to serious signal collisions when there are many surrounding devices. To solve this problem, we propose a novel synchronous communication scheme in this paper, which can greatly improve transmission performance while maintaining low energy consumption. The simulation results effectively verify the effectiveness of the proposed scheme.

It is essential to support real-time communication in networked control systems, such as automotive systems or smart factories. Several studies have been conducted to schedule real-time flows by adjusting their priorities using software-defined networking (SDN). However, because its architecture was not originally designed for realtime communication, unexpected challenges may occur if it is utilized without careful consideration. In particular, because SDN does not provide an atomic method for priority adjustment, some flows may violate their timing constraints by receiving additional interference during priority adjustment. In this paper, we propose a novel scheme to safely update the flow priorities in SDN-based real-time systems. Accordingly, we first analyze the challenge that may occur when adjusting flow priorities within a switch. We then develop a scheme that determines safe procedures for priority adjustment. By network emulations, we demonstrate that the above challenge occurs in 178 out of 300 flow sets, but our scheme can effectively prevent all of them.

Cybersickness is one of the factors that deteriorates user experience in virtual reality (VR). To understand how cybersickness is presented through human reactions and responses, we conducted a user study with 13 participants and built a ResNet-BiLSTM-based model that learns visual factors, eye movement, head movement, and physiological signals. The study results show that the model using all modalities yielded a performance of 0.88 F1-score. In particular, the model using the data that can be collected by HMD (Head Mounted Display) showed 0.87 F1-score, comparable to the model using all modalities, which indicates that cybersickness can be sufficiently well predicted through basic VR equipment (HMD). Finally, we present the importance of individual characteristics in cybersickness modeling.

Software-Defined Networks (SDN) need a reliable and robust control plane. To make the control plane available to the underlying network, the controller must be available all the time. To deal with failures, the switches of the failed controller are assigned to a slave controller. Hence, the slave controller is available in backup to the switches attached with a master controller. In this paper, we formulate the problem of switch migration to the slave controller as a 0/1 knapsack problem. Finally, we compared our scheme with the static method of switch migration. The results show the effectiveness of our proposed scheme.

In light of the recent advancements made in IT, many researchers are studying and exploring ways to minimize damage from fire disasters using artificial intelligence and cloud technology. With the introduction of edge computing, firedisaster response software systems have made significant progress. However, existing studies often do not consider the response to a sudden power supply cut-off due to fire. In this study, we propose a container migration scheme based on the first-fit-decreasing algorithm of bin-packing problem and 0-1 knapsack algorithm to provide fault tolerance for containers running on edge servers that are powered off.

Unsupervised Domain Adaptation of object detection can prevent performance degradation for new environment which does not include annotation We improved the performance by applying Pseudo label.

Probabilistic modeling of normal data is commonly used for anomaly detection. In this paper, we present a novel anomaly detection process using a score-based probabilistic generative model. Our method is based on the Langevin dynamics-based sampling methods, but we use a reverse trajectory of a standard score-based framework to compute anomaly scores. We validate our anomaly detection framework according to different one-class classification settings on the MNIST dataset.

With the recent rapid technological advancement, numerous frameworks and platforms are launched especially for IoT yet, it is difficult in general to integrate a new framework or a platform to a legacy system. Microservices architecture, a popular modern architectural style, makes it easy to apply new technologies by decomposing entire systems into independent microservices. With these advantages, many studies have applied the microservice architecture to IoT. However, most of the studies considered IoT and MSA, excluding MR, a new promising user interface in IoT. This project aims to create a novel interoperable architecture that combines MR and IoT based on MSA.

SDN is a next-generation networking architecture that separates the network into a control plane and a data plane and allows the traffic forwarding to be controlled and managed by a software-based controller, enabling efficient network management. Riverbed Modeler, which is mainly used for tactical network M&S, provides a model for SDN simulation. Recently, a testbed for performing simulations of Riverbed Modeler-based multi-domain SDN has been implemented. However, since the PC is different for each domain, the simulation execution process is cumbersome, and it is difficult to apply traffic information at once. In this paper, we present an extended testbed architecture and implementation results in which one simulator includes a multi-domain environment.

Wi-Fi, a well-established communication technology for various applications of wireless networking, has seen its share of its evolution. In this survey, we briefly explain this trend as well as its current technological improvements in IEEE standardization and activities in Wi-Fi Alliance to understand its direction in the future.

Fire, one of the major disasters that commonly occur in urban areas, frequently results in loss of life and property. Therefore, it is very important and urgent to find effective ways to minimize the loss due to Fire Disaster. Casualties can be minimized if we get the right kind of trainings to respond to such disasters. Over the past decade, training in Virtual Reality (VR) for military and disaster preparedness has been increasingly recognized as an important adjunct to traditional modalities of real-life drills. Virtual Reality based solutions provide immersive and interactive experience training that implies improved learning through an increased amount of presence and at a low cost. Hence, in order to minimize the potential harm from the fire hazards, a rational VR based fire training simulator taking some basic account of the various aspects of fire hazards has been developed and is described herein. In this simulator, a content is designed to create a realistic fire environment for the purposes of effective virtual training, which allows the trainees to experience a realistic and yet non-threatening fire scenario. Although this content design is very primitive in nature and not considering all aspects of fire disaster drill, it allows the trainee to learn the use of fire extinguisher and take some basic safety actions in case of fire.

Measurement of a biomedical signal is representative method to detect whether drivers are drowsy. The biomedical signal fundamentally has oscillating properties, and that are obtained from high-sensitive pressure sensors. Here, we demonstrate a distinguished high-sensitive pressure sensor based on suspended 2D material. An acoustic signal is regarded as remote pressure signal in this work, and those are measured with real-time current variation. The optimized real-time signals are exhibited with a fast Fourier transform analysis and a noise cancellation method. We believe that our approach to the suspended 2D material-based pressure sensor paves the way for detecting biomedical signals well.

Resistive random-access memory (RRAM), one of the most potential candidates for synaptic devices, has been studied steadily for many years. However, destructive switching methods and process variations acted as factors that were difficult to apply to the neuromorphic system. In particular, the breakdown of switching layer may occur even before training is sufficiently performed if endurance is not secured in on-chip training. In this work, we propose a binary neural network of a hardware friendly learning algorithm to overcome this issue at system-level study. Binary neural network (BNN) can accelerate the time at which the recognition rate is saturated because all weight states are defined by one switching event. In addition, the resistance to variation can be improved by using the maximum/minimum of the current level of the memristors. However, the conventional BNN has the disadvantage that batch normalization and real value weights must be used together for learning. In this paper, we verified a method for learning BNN using boundary values.

Due to the recent improvement in the computational power of hardware and the growth of data, a deep learning-based approach that optimizes parameters using massive data showed excellent performance. In computer vision, research using a convolutional neural network(CNN) is being actively conducted. However, it is challenging to apply to IoT devices due to the high computational complexity and massive memory usage required. In this paper, we propose a quantized CNN hardware for IoT devices that optimized memory usage and computation complexity. In addition, we present a quantization framework for the proposed hardware design. The presented framework includes floating-point training, quantization, fully integer arithmetic inference, and hardware design processes. As a result of implementing the quantized CNN on the Xilinx ZC702 evaluation board, power consumption and inference speed improved by 4.86× and 2.58×, respectively, compared to 32-bit floating-point hardware.

Binarized neural network (BNN) is one of the most efficient neural network for low-cost convolution operations. In BNN, binarized data is utilized to reduce memory size and complexity of convolution operations. A content addressable memory (CAM) based BNN accelerator can perform convolution operations efficiently by taking an advantage of fully parallel search operations in CAM. However, one of the critical issue of CAM based BNN hardware is that the operation reliability is severely degraded by the process variation during ML sensing operation. Therefore, we propose new CAM array design which can reduce hardware error probability. The proposed CAM based accelerator achieves 62% reduction in XNOR-popcount operations, and the classification accuracy drop of Fashion MNIST data set reduces from 2.33% to 1.26%.

As electric vehicles’ battery voltage is increased from 400 V to 800 V, low-voltage DC/DC converter (LDC) is required to regulate low output voltage for electronic loads over the high and wide input voltage range of 350–800 V. The entire input voltage range can be divided into two: low and high input voltage ranges. This paper proposes the integrated DC/DC converter, which has two operational modes in these ranges. The proposed converter operates as a three-switch double-ended active clamp forward converter in the low input voltage range and an asymmetric half-bridge converter in the high input voltage range. With two operational modes, the proposed converter can have reductions of the voltage stresses on the switches and DC offset current of the transformer compared with the conventional converters. Thus, it can achieve high efficiency under the entire input voltage range. The proposed converter is verified by a 350–800 V input and 13.9 V/600 W output laboratory prototype.

In this paper, a rotating-gate field-effect transistor has been developed to be used as a self-powered sensor. The developed device showed the drain current increased according to the rotation velocity of the gate motor and saturated under a sufficient drain voltage.

This paper proposes a design method for a driving simulator and driver assistance system in a vehicle network environment. Each module is on the same Controller Area Network(CAN) bus line for vehicle network configuration. To compose hardware in the loop simulation (HILS), a testbed similar to the actual vehicle environment is configured, and a virtual city and vehicle objects are created using a driving simulator. In particular, the vehicle network was configured by modularity and CAN ID applied from the modules according to the priority order. In the experiment, the possibility of using the vehicle control module was confirmed through the vehicle driving scenario in terms of the driver’s various emotions and situations. By acquisition of bio signal data from attached electrodes on the driver’s hand, the response time for changing the vehicle control mode and the control response time was measured in the manual driving mode, and it was confirmed that the control latency time is less than the commercialized response time of 830ms in autonomous vehicle control.