Bongjae Kim

Dynamic Offloading Algorithm for Drone Computation

In this paper, we propose a mobility-aware offloading decision method for tracking moving object on drone. It aims to shorten response time because the time constraint is a crucial criteria for offloading decision making for tasks involved in moving object recognition and tracking. Through the dwell time model of target moving object, we investigate the impact of target mobility and derive a dynamic offloading decision method considering the expected dwell time. Our proposed model is also designed to give feedback on the network failure rate so that it can estimate accurately the response time.

Dynamic QoS Scheme for InfiniBand-Based Clusters

Cluster-based computing systems are very widely used in various fields, including simulations and big data processing. InfiniBand is de-facto interconnect technology for cluster-based computing. QoS is a very important issue in data communication of cluster-based computing systems. In this paper, we propose dynamic QoS scheme for InfiniBand-based clusters. The proposed scheme can change QoS level in terms of the bandwidth. The proposed QoS scheme can get more bandwidth by changing the QoS level. The prototype of the proposed scheme was implemented in a real InfiniBand-based clusters. By using the prototype implemented, we confirmed and evaluated the usefulness and effectiveness of the proposed scheme.

Deep-learning based web UI automatic programming

The GUI building is an important part of web application development. Various studies such as WYSWYG web editor have been conducted to make this job convenient, where the job is composed of sketching of GUI and coding of HTML/CSS from the sketch. In this paper, we propose a novel way of web GUI building with computer vision and deep-learning. The proposed method requires only a hand-drawn sketch to build GUI. It recognizes web layout using computer vision algorithm, and web widgets using Faster R-CNN. With the recognized information, it makes HTML code automatically.

A communication model based offloading decision for flying ad-hoc networks

Flying ad-hoc networks is composed of many connected drones with wireless communication. A drone sends its task to the cloud to reduce task completion time and energy consumption. However, flying ad-hoc networks, where several drones are connected to each other, can divide a task into small tasks and assign each small task to drones to improve responsibility. In this paper, we propose an offloading decision scheme that considers task completion time and energy consumption under the typical wireless communication model. The proposed scheme compares the cost of executing small tasks on the drones with the cost of committing a task to the cloud and decides offloading a task only if the cost of offloading is definitely smaller than the cost of using multiple drones. Our simulation results show that the proposed decision scheme is necessary because offloading spends more energy and time in some cases.

Role-based automatic programming framework for interworking a drone and wireless sensor networks

Wireless sensor networks extends their applications as related technologies are developed. Macro-programming is a centralized approach to sensor network programming and well-performed simple applications. However, as wireless sensor networks work with external entities such as drones, application is more complex. Macro-programming is not suitable to these complex applications. In this paper, we proposed a role-based automatic programming framework that is well-performed even under complex working conditions. The proposed framework running on TinyOS provides the role assignment and transition mechanisms for various roles are working concurrently. We also implemented a prototype to verify availability of the proposed framework.

Performance Evaluations of Multiple GPUs based on MPI Environments

GPU-based computations are widely used in various computing areas because GPU provides very high computing performance when compared to typical CPU. In this paper, we evaluate and analyze the computing performance of multiple GPUs based on MPI environments. We examine the performance of sparse matric-vector multiply (SpMV). SpMV is one of the most heavily used components in many scientific applications. Based on the performance evaluation results, generally, the execution time of SpMV is decreased as the number of GPUs increase. In some case, the performance was reduced according to the computation overhead, the memory copy overhead among GPUs, and the characteristics of sparse matrices.

An energy efficient rendezvous node selection approach

Using a drone as a mobile sink is an interesting issue in wireless sensor networks. There are many studies that try to minimize the flight distance of the drone but their approaches causes unbalanced energy consumption of nodes. In this paper, we propose a new rendezvous node selection approach to achieve the balance of each nodes energy consumption and to minimize the flight distance of the drones. Simulation results show that our approach can extend the entire network life-time and maintain the stable variation of the flight distance.

Design and Implementation of a Wearable Device for the Blind by Using Deep Learning Based Object Recognition

Recently, deep learning based object recognition systems are very widely used in various fields, including surveillance systems. The accuracy of object recognition based on deep learning is better than other schemes. In this paper, we propose a wearable device for the blind by using deep learning based object recognition. Based on the implemented prototype and evaluation results, we confirmed the usefulness and effectiveness of the proposed wearable device.

IoT-Based VR Service Model to Improve Exercise Capacity

Recently, Internet of Thing has been actively studied in various fields such as wearable devices, smart cars, and smart factories. According to the Gartner report of 2017, the IoT market is expected to grow by more than

Implementation of Recommender System Based on Personalized Search Using Intimacy in SNS

2 trillion by 2020. In this paper, we aim to develop a VR service model that combines healthcare and virtual reality through IoT technology to improve exercise capacity. In other words, we propose VR service model that can improve the exercise needs of the user by using bicycle which can be easily accessed by people for improving the IoT based athletic performance and incorporating VR service. The proposed model improves the sense of reality through 3D modeling using virtual reality technology, and the user can intuitively confirm the driving record and driving information, and can increase the efficiency of the user’s motion through the target heart rate.