• ETRI Journal
• /
• v.40 no.4
• /
• pp.421-434
• /
• 2018

The application of deep neural networks (DNNs) to connect the world with cyber physical systems (CPSs) has attracted much attention. However, DNNs require a large amount of memory and computational cost, which hinders their use in the relatively low-end smart devices that are widely used in CPSs. In this paper, we aim to determine whether DNNs can be efficiently deployed and operated in low-end smart devices. To do this, we develop a method to reduce the memory requirement of DNNs and increase the inference speed, while maintaining the performance (for example, accuracy) close to the original level. The parameters of DNNs are decomposed using a hybrid of canonical polyadic-singular value decomposition, approximated using a tensor power method, and fine-tuned by performing iterative one-shot hybrid fine-tuning to recover from a decreased accuracy. In this study, we evaluate our method on frequently used networks. We also present results from extensive experiments on the effects of several fine-tuning methods, the importance of iterative fine-tuning, and decomposition techniques. We demonstrate the effectiveness of the proposed method by deploying compressed networks in smartphones.

• Journal of Biomedical Engineering Research
• /
• v.26 no.5
• /
• pp.331-335
• /
• 2005

The fusion technology of small sensor and wireless communication was followed by various application examples of the embedded system, where the social infrastructural facilities and ecological environment were wirelessly monitored. In addition, this technology represents the primary application area being extended into the healthcare field. In this study, a body area network for ubiquitous healthcare is presented. More specifically this represents a wireless biomedical signal acquisition device characterized by small size, low power consumption, pre-processing and archiving capability. Using this device, a new method for monitoring vital signs and activity is created. A PDA-based wireless sensor network enables patients to be monitored during their daily living, without any constraints. Therefore, the proposed method can be used to develop Activities of Daily Living (ADL) monitoring devices for the elderly or movement impaired people. A medical center would be able to remotely monitor the current state of elderly people and support first-aid in emergency cases. In addition, this method will reduce medical costs in society, where the average life expectancy is increasing.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.2346-2349
• /
• 2005

Distributed Autonomous Robotic System (DARS) is defined as a system that independent autonomous robots in the restricted environments infer their status from pre-assigned conditions and operate their jobs through the cooperation with each other. In the DARS, a robot contains sensor part to percept the situation around themselves, communication part to exchange information, and actuator part to do a work. Especially, in order to cooperate with other robots, communicating with other robots is one of the essential elements. Because Bluetooth has many advantages such as low power consumption, small size module package, and various standard protocols, Bluetooth is rated as one of the efficient communicating technologies which can apply to small-sized robot system. In this paper, we will develop Bluetooth communicating system for autonomous robots such as DARS robots. For this purpose, The Bluetooth communication system must have several features. The first, this system should be separated from other robot parts and operate spontaneously and independently. In other words, this communication system should have the ability to organize and maintain and reorganize a network scheme. The next, this system had better support any kinds of standard interfaces in order to guarantee flexible applicability to other embedded system. We will discuss how to construct and what kind of procedure to develop the network system.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.10 no.5
• /
• pp.293-298
• /
• 2010

The key of USN(Ubiquitous Sensor Network) technology is low power wireless communication technology and proper resource allocation technology for efficient routing. The distinguished resource allocation method is needed for efficient routing in sensor network. To solve this problems, we propose an algorithm that can be adopted in USN with making up for weak points of PQ and WRR in this paper. The proposed algorithm produces the control discipline by the fuzzy theory to dynamically assign the weight of WRR scheduler with checking the Queue status of each class in sensor network. From simulation results, the proposed algorithm improves the packet loss rate of the EF class traffic to 6.5% by comparison with WRR scheduling method and that of the AF4 class traffic to 45% by comparison with PQ scheduling method.

• Proceedings of the KIEE Conference
• /
• 2006.04a
• /
• pp.177-179
• /
• 2006

This paper presents an analysis architecture of embedded operating systems for wireless sensor network. Wireless multi-hop sensor networks use battery-operated computing and sensing device. We expect sensor networks to be deployed in an ad hoc fashion, with very high energy constraints. These characteristics of multi-hop wireless sensor networks and applications motivate an operating system that is different from traditional embedded operating system. These days new wireless sensor network embedded operating system come out with some advances compared with previous ones. The analysis is focusing on understanding differences of dominant wireless sensor network OS, such as TinyOS 2.0 with TinyOS 1.x.

• Journal of the Korean Society of Radiology
• /
• v.10 no.8
• /
• pp.637-643
• /
• 2016

Due to the popularity of artificial intelligent, medical image processing using artificial neural network is increasingly attracting the attention of academic and industry researches. Deep learning with a convolutional neural network has been proved to very effective representation of images. However, the training process requires high performance H/W platform. Thus, the realtime learning of a large number of high dimensional samples within low-power devices is a challenging problem. In this paper, we attempt to establish this possibility by presenting a realtime neural network method on Raspberry pi using online sequential extreme learning machine. Our experiments on high-dimensional dataset show that the proposed method records an almost real-time execution.

• The KIPS Transactions:PartC
• /
• v.13C no.3 s.106
• /
• pp.375-382
• /
• 2006

As various kinds of embedded systems (or devices) become widely available, research on home network middleware which can access and control embedded home appliances are actively being progressed. However, there is a significant problem in applying the home network technology to embedded systems because of their limited storage space and low computing power. In this paper, we present a lightweight middleware for home network on embedded systems. Also, we propose a mechanism for mobility management which adopts the anycast technology.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39B no.11
• /
• pp.771-778
• /
• 2014

We propose Femto-cell power control scheme in HeNet with Game Theory. The Femto-cell which provide high quality with low power is issued by many benefits, however there is a bunch of interferences when many Femto-cells use overlapped bandwidth with Macro-cell. We defined base station of cellular networks and mobile users as players of Game Theory, and configured interference effect among each other as power utility function. Futhermore, we showed enhanced overall system performance, lower power usage and interference decrease by using optimal power.

• Proceedings of the KIPE Conference
• /
• 2004.07b
• /
• pp.733-737
• /
• 2004

오랫동안 연구되어 오던 전력선 통신 기술은 1990 년대 중반에 이르러, 모뎀 기술을 응용한 10kbps급의 저속 전력선통신 기술이 개발되어 상용화 수준에 근접한 제품이 등장하기 시작했다. 반면 고속 전력선통신 기술 분야는 관심을 갖기 시작한 시기가 저속 분야에 비해 상대적으로 늦지만, 국내외 여러업체에서 고속 전력선통신 솔루션을 개발, 상용화에 근접한 칩셋 및 제품들을 선보이고 있다. 국내에서는 산업자원부의 중기거점 사업으로 수행중인 "50Mbps급 고속 전력선 통신망 개발사업"과 관련하여 고속 전력선 통신망 장치를 실용화 할 수 있도록 집단주택과 단독주택 형태의 인터넷 시범사업에 관한 연구가 활발히 수행중이다. 본 연구에서는 실제 수용가 환경인 단독주택 약 20세대에서 수행한 인터넷 시범사업을 통하여 개발된 전력선 통신 모뎀과 시스템에 대한 성능 측정을 수행한다. 이를 토대로 현 단계에서의 기술을 분석하고, 전력선 통신망에 대한 운영상의 적정 노드 및 가입자망 전송 섹터범위 등의 종합적인 평가에 대한 내용 등 실증시험망 구축 및 운용에 대한 결과와 향후 활용방안에 대해서 기술하고자 한다.

• Proceedings of the KIEE Conference
• /
• 2015.07a
• /
• pp.335-336
• /
• 2015

Voltage collapse is one of the main threats in power system operation and it is a result of a very low voltage profile in one or multiple areas in the power system network. Hence, the need for proper coordination among transmission and generator operators to maintain voltage stability if of paramount importance. In this paper, the different voltage and reactive power operating guidelines and criteria of various electric reliability councils, coordinating councils, transmission system operators and other electric utilities are discussed. Among these are NERC, ENTSO-E, WECC, PJM and ERCOT. A comprehensive comparison is also made to the existing Korean power system voltage reliability guidelines. Furthermore, this paper presents a set of voltage and reactive power guidelines that can help in determining voltage limits that can improve the reliability of the Korean power system operation.