• Journal of Internet of Things and Convergence
• /
• v.10 no.2
• /
• pp.103-108
• /
• 2024

In this paper, a smart elevator system was studied using real-time object detection technology based on YOLO(You only look once)v5. When an external elevator button is pressed, the YOLOv5 model analyzes the camera video to determine whether there are people waiting, and if it determines that there are no people waiting, the button is automatically canceled. The study introduces an effective method of implementing object detection and communication technology through YOLOv5 and MQTT (Message Queuing Telemetry Transport) used in the Internet of Things. And using this, we implemented a smart elevator system that determines in real time whether there are people waiting. The proposed system can play the role of CCTV (closed-circuit television) while reducing unnecessary power consumption. Therefore, the proposed smart elevator system is expected to contribute to safety and security issues.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.3
• /
• pp.107-116
• /
• 2024

In this paper, we analyze the performance of MicroVM when running AI applications on an edge computing environment and whether it can replace current container technology and traditional virtual machines. To achieve this, we set up Docker container, Firecracker MicroVM and KVM virtual machine environments on a Raspberry Pi 4 and executed representative AI applications in each environment. We analyze the inference time, total CPU usage and trends over time and file I/O performance on each environment. The results show that there is no significant performance difference between MicroVM and container when running AI applications. Moreover, on average, a stable inference time over multiple trials was observed on MicroVM. Therefore, we can confirm that executing AI applications using MicroVM instead of container or heavy-weight virtual machine is suitable for an edge computing.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.5
• /
• pp.787-799
• /
• 2023

As IoT devices are widely used at home, IoT automation system that is integrate IoT devices for users' demand are gaining populrity. There is automation rule in IoT automation system that is collecting event and command action. But attacker delay the packet and make time that real state is inconsistent with state recongnized by the system. During the time, the system does not work correctly by predefined automation rule. There is proposed some detection method for delay attack, they have limitations for application to IoT systems that are sensitive to traffic volume and battery consumption. This paper proposes a practical packet delay attack detection technique that can be applied to IoT systems. The proposal scheme in this paper can recognize that, for example, when a sensor transmits an message, an broadcast packet notifying the transmission of a message is sent to the Server recognized that event has occurred. For evaluation purposes, an IoT system implemented using Raspberry Pi was configured, and it was demonstrated that the system can detect packet delay attacks within an average of 2.2 sec. The experimental results showed a power consumption Overhead of an average of 2.5 mA per second and a traffic Overhead of 15%. We demonstrate that our method can detect delay attack efficiently compared to preciously proposed method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.41 no.11
• /
• pp.1589-1597
• /
• 2016

In this paper, we propose a system in a large-scale environment, such as desert and ocean, that can reduce the overall transmission power consumption in mobile sensor network. It is known that the transmission power consumption in wireless sensor network is proportional to the square of transmission distance. Therefore, if the locations of mobile sensors are far from the sink node, the power consumption required for data transmission increases, leading to shortened operating time of the sensors. Hence, in this paper, we propose a system that can reduce the power consumption by allowing to transmit data only if the transmission range of the sensors is within a predetermined distance. Moreover, the energy efficiency of the overall sensor network can even be improved by reducing the number of data transmissions at the sink node to gateway based on compressed sensing. The proposed system is actually implemented using Arduino and Raspberry Pi and it is confirmed that source data can be approximately decoded even when the gateway received encoded data fewer than the required number of data from the sink node. The performance of the proposed system is analyzed in theory.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.7
• /
• pp.446-452
• /
• 2020

As the use of children's school buses increases, accidents caused by the negligence of school bus drivers and ride carers have also increased significantly. To prevent such accidents, the government is coming up with various policies. We propose an accident prevention and safety management system for children's school buses. Through this system, bus drivers can easily check whether each child is seated and whether the seat belt is used, so it is possible to quickly respond to children's conditions while driving. With the ability to recognize faces by analyzing camera images, children can use a seat belt that is automatically adjusted to their height. It is therefore possible to prevent secondary injuries that may occur in the event of a traffic accident. In addition, a sleeping child-check system is provided to confirm that all children get off the bus, and a text service is provided to inform parents of their children's locations in real time. Based on Raspberry Pi, the system is implemented with cameras, pressure sensors, motors, Bluetooth modules, and so on. This proposed system was attached to a bus model to confirm that the series of functions work correctly.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.5
• /
• pp.563-569
• /
• 2018

This paper is on the implementation of a system for automatically monitoring window farm hydroponics based on Arduino (utilizing Arduino's open source code) emerging as the icon of the Fourth Industrial Revolution. A window farm, which means window-type hydroponics, is offered as an alternative to fulfill the desires of people who want to grow plants aside from the busy daily life in the city. The system proposed in this paper was developed to automatically monitor a window farm hydroponics cultivation environment using the Arduino UNO board, a four-charmel motor shield, temperature and humidity sensors, illumination sensors, and a real-time clock module. Modules for hydroponics have been developed in various forms, but power consumption is high because most of them use general power and motors. Since it is not a system that is monitored automatically, there is a disadvantage in that an administrator always has to manage its operational state. The system is equipped with a water supply that is most suitable for a plant growth environment by utilizing temperature, humidity, and light sensors, which function as Internet of Things sensors. In addition, the real-time clock module can be used to provide a more appropriate water supply. The system was implemented with sketch code in a Linux environment using Raspberry Pi 3 and Arduino UNO.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2018.05a
• /
• pp.347-350
• /
• 2018

In order for an IoT system to automatically make the house temperature pleasant for the user, the system needs to predict the optimal start-up time of air-conditioner or heater to get to the temperature that the user has set. Predicting the optimal start-up time is important because it prevents extra fee from the unnecessary operation of the air-conditioner and heater. This paper introduces an ANN(Artificial Neural Network) and an IoT system that predicts the cooling and heating time in households using air-conditioner and heater. Many variables such as house structure, house size, and external weather condition affect the cooling and heating. Out of the many variables, measurable variables such as house temperature, house humidity, outdoor temperature, outdoor humidity, wind speed, wind direction, and wind chill was used to create training data for constructing the model. After constructing the ANN model, an IoT system that uses the model was developed. The IoT system comprises of a main system powered by Raspberry Pi 3 and a mobile application powered by Android. The mobile's GPS sensor and an developed feature used to predict user's return.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.18 no.1
• /
• pp.135-141
• /
• 2018

In this paper, we describe design and implementation of optimal smart home control system. Recent developments in technologies such as sensors and communication have enabled the Internet of Things to control a wide range of objects, such as light bulbs, socket-outlet, or clothing. Many businesses rely on the launch of collaborative services between them. However, traditional IoT systems often support a single protocol, although data is transmitted across multiple protocols for end-to-end devices. In addition, depending on the manufacturer of the Internet of things, there is a dedicated application and it has a high degree of complexity in registering and controlling different IoT devices for the internet of things. ARIoT system, special marking points and edge extraction techniques are used to detect objects, but there are relatively low deviations depending on the sampling data. The proposed system implements an IoT gateway of object based on OneM2M to compensate for existing problems. It supports diverse protocols of end to end devices and supported them with a single application. In addition, devices were learned by using deep learning in the artificial intelligence field and improved object recognition of existing systems by inference and detection, reducing the deviation of recognition rates.

• Journal of Wetlands Research
• /
• v.22 no.4
• /
• pp.312-320
• /
• 2020

Various LIDs with natural water circulation function are applied to reduce urban environmental problems and environmental impact of development projects. However, excessive Infiltration and evaporation of LID facilities dry the LID internal soil, thus reducing plant and microbial activity and reducing environmental re duction ability. The purpose of this study was to develop a real-time measurement system with complex sensors to derive the management plan of LID facilities. The test of measurable sensors and Internet of Things (IoT) application was conducted in artificial wetlands shaped in acrylic boxes. The applied sensors were intended to be built at a low cost considering the distributed LID and were based on Arduino and Raspberry Pi, which are relatively inexpensive and commercialized. In addition, the goal was to develop complex sensor measurements to analyze the current state o f LID facilities and the effects of maintenance and abnormal weather conditions. Sensors are required to measure wind direction, wind speed, rainfall, carbon dioxide, Micro-dust, temperature and humidity, acidity, and location information in real time. Data collection devices, storage server programs, and operation programs for PC and mobile devices were developed to collect, transmit and check the results of measured data from applied sensors. The measurements obtained through each sensor are passed through the Wifi module to the management server and stored on the database server in real time. Analysis of the four-month measurement result values conducted in this study confirmed the stability and applicability of ICT technology application to LID facilities. Real-time measured values are found to be able to utilize big data to evaluate the functions of LID facilities and derive maintenance measures.