• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.7B
• /
• pp.694-702
• /
• 2009

Geographical routing protocol, where nodes only need local information exchange to make routing decisions, is a very efficient routing scheme for wireless sensor networks in scalability. However, pure geographical routing does not take account of wireless link condition. If wireless link condition is not considered, when the routing table is updated, the nodes with bad link conditions are updated in the routing table and can be chosen as the next hop. This brings out the retransmissions because of received packet's errors. Also, because of these retransmissions, additional power is consumed. In this paper, we propose geographical routing scheme considering wireless link condition, where reliable data transmission is made and the consumed energy is minimized.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2014.11a
• /
• pp.30-32
• /
• 2014

최근 낸드 플래시 메모리는 임베디드 저장 장치로 많이 사용되고 있다. 비휘발성인 플래시 메모리는 기존의 하드디스크와 달리 저 전력, 좋은 내충격성 및 집적도 등 많은 장점이 있지만 데이터 업데이트 시 덮어쓰기가 안 되어 쓰기 연산 전 해당 블록을 지우는 작업이 선 진행되어야 하며 이로 인해 부분 페이지 업데이트가 자주 일어난다. 이런 플래시메모리와 더불어 최근 차세대 메모리연구가 많이 진행 중인데, 이 중에서 PCM 이라는 메모리는 비휘발성으로 정전 시 데이터가 날라 가버리는 DRAM에 반해 전원이 공급 안 되더라도 데이터가 보존되는 특성이 있다. 하지만 PCM 역시 플래시 메모리와 마찬가지로 블록 당 쓰기연산 작업이 제한되어 있어서 근래에 DRAM과 같이 사용하는 하이브리드 구조를 채택하여 많은 연구가 진행되고 있다. 따라서 본 논문에서는 플래시 메모리의 문제점을 해결함으로서 수명을 연장시키고 정 전시 데이터가 보존되지 않는 DRAM의 단점을 하이브리드 메모리를 기반으로하여 데이터의 신뢰성을 높이는 FTL알고리즘을 제안한다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.10a
• /
• pp.569-572
• /
• 2016

It is currently associated with the increase in smart home devices. So they are getting a lot IoT platform for managing and integrating these devices. However, existing platforms lack the firmware management for each device. In this paper, We will make easier the management of firmware, using a technique called POTA. This technology is a technology that allows you to update the firmware using a wireless internet without a computer and Utilizing the TR-069 protocol, we propose a system for controlling each device in CPE.

• Journal of Korea Game Society
• /
• v.19 no.4
• /
• pp.85-96
• /
• 2019

In this note, we propose a game system to fairly and transparently control and update game variables using blockchain. As an example, we present an Ethereum-based maze escape game, being the first blockchain-based maze game and the first attempt to control game level through blockchain. In this system, the level function computes maze level reflecting the amount of maze walls purchased/sold by users. Using blockchain for game management saves human and economic resources and prevents unfair practice of game developers unilaterally manipulating game level or item enhancement rate.

• Journal of Korea Society of Industrial Information Systems
• /
• v.15 no.3
• /
• pp.67-75
• /
• 2010

Network reprogramming is a technology that allows several sensor nodes deployed in sensor field to be repaired remotely. Unlike general communication in sensor network where small amount of data is transferred, network reprogramming requires reliable transfer of large amount of data. The existing network reprogramming techniques suffers high cost and large energy consumption to recover data loss in node communication. In this paper, a cluster based network reporgramming scheme is proposed for sensor network. It divides sensor field into several clusters and chooses a cluster header in charge of data relay to minimize duplicated transmission and unnecessary competition. It increases reliability by effective error recovery through status table.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.597-599
• /
• 2022

When IoT device performs on-device AI, the device is required to use various AI models selectively according to target service and surrounding environment. Also, AI model can be updated by additional training such as federated learning or adapting the improved technique. Hence, for successful on-device AI, IoT device should acquire various AI models selectively or update previous AI model to new one. In this paper, we propose AI model repository to tackle this issue. The repository supports AI model registration, searching, management, and deployment along with dashboard for practical usage. We implemented it using Node.js and Vue.js to verify it works well.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.683-689
• /
• 2023

자연어 처리 분야는 최근에 큰 발전을 보였으며, 특히 초대규모 언어 모델의 등장은 이 분야에 큰 영향을 미쳤다. GPT와 같은 모델은 다양한 NLP 작업에서 높은 성능을 보이고 있으며, 특히 챗봇 분야에서 중요하게 다루어지고 있다. 하지만, 이러한 모델에도 여러 한계와 문제점이 있으며, 그 중 하나는 모델이 기대하지 않은 결과를 생성하는 것이다. 이를 해결하기 위한 다양한 방법 중, Retrieval-Augmented Generation(RAG) 방법이 주목받았다. 이 논문에서는 지식베이스와의 통합을 통한 도메인 특화형 질의응답 시스템의 효율성 개선 방안과 벡터 데이터 베이스의 수정을 통한 챗봇 답변 수정 및 업데이트 방안을 제안한다. 본 논문의 주요 기여는 다음과 같다: 1) QA Pair Passage RAG을 활용한 새로운 RAG 시스템 제안 및 성능 향상 분석 2) 기존의 LLM 및 RAG 시스템의 성능 측정 및 한계점 제시 3) RDBMS 기반의 벡터 검색 및 업데이트를 활용한 챗봇 제어 방법론 제안

• The KIPS Transactions:PartC
• /
• v.11C no.7 s.96
• /
• pp.917-922
• /
• 2004

Mobile ad hoc networks (MANETs) are networks of mobile nodes that have no fixed network infrastructure. Since the mobile node's location changes frequently, it is an attractive area to maintain the node's location efficiently. In this paper, we present an adaptive randomized database group (ARDG) scheme to manage the mobile nodes mobility in MHANETs. The proposed scheme stores the network nodes' location in location databases to manage the nodes' mobility. When a mobile node changes its location or needs a node's location, the node randomly select some databases to update or que교 the location information. The number of the selected databases is fixed in the case of querying while the number of the databases is determined according to the node's popularity in the case of updating. We evaluated the performance of the proposed scheme using an analytical model, and compared the performance with that of the conventional randomized database group (RDG) scheme.

• Journal of IKEEE
• /
• v.19 no.4
• /
• pp.583-589
• /
• 2015

In the HEVC CABAC, the probability model is updated after a bin is encoded and next bin is encoded based on the updated probability model. Conventional CABAC encoders can encode only one bin per cycle, which cannot increase the encoding throughput. The probability model does not need to be updated in the bypass bins. In this paper, a HEVC CABAC encoder is proposed to increase encoding throughput by parallel processing of bypass bins. The designed CABAC encoder can process either a regular bin or maximum 4 bypass bins in a cycle. On the average, it can process 1.15~1.92 bins in a cycle. Synthesized in 0.18 um technology, its gate count, maximum operating speed, and the maximum throughput are 78,698 gates, 136 MHz, and 261 Mbin/s, respectively.

• Journal of Software Engineering Society
• /
• v.28 no.1
• /
• pp.1-6
• /
• 2019

Google released an initial version of Android that runs Dex(Dalvik Executable) through the Dalvik Runtime. Since Dalvik Runtime is based on interpreter, JIT(Just-in-time) compilation has been applied to improve performance. After Lollipop(Android 5.0) Dalvik Runtime has replaced with ART Runtime which support AOT (Ahead-of-time) compilation of Dex into Native Code. The late st Android has a problem that the application execution speed is slow until the AOT compilation is completed according to the actual usage record after the installation of the app. To improve the problem we have investigate the characteristics of profile that can improve the execution speed of the application and generate the profile automatically. Finally we propose a method that can optimize the application at install time. With the proposed method we can optimize selectively at install time and can help improving the execution speed of the app from the initial execution.