• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04d
• /
• pp.163-165
• /
• 2003

P2P 시스템의 주요 이슈는 자원 (resource) 을 효율적으로 저장하고 찾는 것이다. 자원 탐색 프로토콜은 초기의 Napster, Gnutella와 같은 형태에서 발전하여, 현재는 분산 해시 테이블 (Distributed Hash Table을 사용한 형태로 발전하고 있다. 본 논문에서는 분산 해시 테이블을 사용한 P2P 프로토콜인 Magic Square를 제안한다. Magic Square에 참여하는 칵 노드는 양방향 스킬리스트로 구성된 지역 라우팅 테이블과 임의의 노드로 구성된 전역 라우팅 테이블을 가진다. 지역 라우팅 테이블은 각 노드의 능력을 고려하여 구성된다. 스킵리스트를 사용하였기 때문에 탐색과 노드의 추가라 삭제 과정이 간단하며. P2P 네트워크가 자주 바뀌어도 큰 영향을 받지 않는다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.427-429
• /
• 2023

인공지능 기술의 발전에 따라 유체 시뮬레이션 분야에서는 복잡한 액체의 흐름을 모사하기 위해 기계학습 기술이 많이 활용되고 있다. 이러한 시뮬레이션에서 성능 향상의 가장 중요한 요소는 학습 데이터다. 이 논문에서는 기계학습 기반 유체 시뮬레이션의 학습 데이터 생성 단계 중 기존의 방법보다 효율적으로 비말(splash) 탐색하는 방법을 제안한다. 기존 방법에서는 CPU 환경에서 큐(queue)를 이용하는 너비우선탐색(breadth first search) 기법을 사용하기 때문에 처리속도가 느리다. 반면에 제안하는 기법에서는 배열로 되어 있는 해시 테이블(hash table)을 이용해 충돌 문제를 해결해 GPU 환경에서 비말을 신속하게 검출하도록 하기 때문에 빠른 학습 데이터 생성이 가능하도록 했다. 이 알고리즘의 유효성을 확인하기 위하여 정확성과 수행시간을 확인하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2007.11a
• /
• pp.982-985
• /
• 2007

인터넷의 보급과 PC의 발달로 정보의 분산과 공유가 가속화를 이루었고, 컴퓨팅 구조가 인터넷 기반으로 변화하였다. 이후 지금까지 수많은 시스템들이 클라이언트/서버 모델을 유지하고 있는 한편 대단위 분산 다중서버로 구성되는 환경이 출현하였다. 클라이언트/서버 모델에서 사용자들의 정보 공유의 제약을 해소하기 위해 개인의 피어 사이에 서로 공유하는 P2P(Peer-to-Peer) 패러다임이 출현 되었다. 하지만 기존의 DHT(Distributed Hash Table)를 기반으로 하는 P2P 시스템에선 정보를 검색할 때, 검색어와 검색결과의 모호성이 존재하고, 범위 검색을 지원하지 못하는 문제가 있었다. 우리는 이 문제를 해결할 한 가지 방안을 본 논문에서 제안한다.

• Journal of Information Technology Services
• /
• v.8 no.3
• /
• pp.171-184
• /
• 2009

Since the development of the Distributed Hash Tables (DHTs), the distributed lookup services are one of the hot topics in the networking area. The main reason of this popularity is the simplicity of the lookup structure. However, the simple key based search mechanism makes the so called "keyword" based search difficult if not impossible. Thus, the applicability of the DHTs is limited to certain areas. In this paper. we find that DHTs can be used as the ubiquitous sensor network (USN) metadata lookup service across a large number of sensor networks. The popularity of the Ubiquitous Sensor Network has motivated the development of the USN middleware services for the sensor networks. One of the key functionalities of the USN middleware service is the lookup of the USN metadata, by which users get various information about the sensor network such as the type of the sensor networks and/or nodes, the residual of the batteries, the type of the sensor nodes. Traditional distributed hash table based lookup systems are good for one sensor network. However, as the number of sensor network increases, the need to integrate the lookup services of many autonomous sensor networks so that they can provide the users an integrated view of the entire sensor network. In this paper, we provide a hybrid lookup model, in which the autonomous lookup services are combined together and provide seamless services across the boundary of a single lookup services. We show that the hybrid model can provide far better lookup performance than a single lookup system.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.236-239
• /
• 2021

With the growth of AI technology, the scale of Knowledge Graphs continues to be expanded. Knowledge Graphs are mainly expressed as RDF representations that consist of connected triples. Many RDF storages compress and transform RDF triples into the condensed IDs. However, if we try to transform a large scale of RDF triples, it occurs the high processing time and memory overhead because it needs to search the large ID mapping table. In this paper, we propose the method of converting RDF triples using Hash-based ID mapping tables with MapReduce, which is the software framework with a parallel, distributed algorithm. Our proposed method not only transforms RDF triples into Integer-based IDs, but also improves the conversion speed and memory overhead. As a result of our experiment with the proposed method for LUBM, the size of the dataset is reduced by about 3.8 times and the conversion time was spent about 106 seconds.

• Convergence Security Journal
• /
• v.22 no.4
• /
• pp.25-34
• /
• 2022

This paper proposes a secret sharing based fast node authentication technique applicable to Flying Ad-Hoc Network (FANET) that can be used to construct self-organized communication network in multi drones and drone squadrons operations. Before deployment, each node stores an exponential share, exponential secret and a portion of PUF CRP table. After being deployed in the field, in the early-stage of network formation, each node broadcasts its ID, exponential share and a hash value of PUF Response and pseudo-random number. Then each node performs a reconstruction of the exponential secret using the exponential shares transmitted from neighboring nodes. When the exponential secret is reconstructed, simultaneous authentication is completed for all nodes that have transmitted the exponential share used in the reconstruction. A node that transmits an incorrect exponential share to disturb the reconstruction of the exponential secret during the authentication process can be detected before performing the reconstruction through the verification of the hash value, and will be excluded from the reconstruction.

• Journal of KIISE:Information Networking
• /
• v.31 no.6
• /
• pp.643-658
• /
• 2004

Mobile ad hoc networks(MANET) have recently attrarted a lot of attention in the research community as well as in industry. Although the previous research mainly focused on the various problems of MANET in data link and network layers, we consider, in this paper, how to efficiently support applications such as service discovery on top of MANET. Peer-to-Peer(P2P) overlay network can be adopted to service discovery mechanism because P2P and MANET share certain similarities, primarily the fact that both arc instances of self-organizing decentralized systems. Especially, distributed hash table(DHT) systems used for r2r overlay network can be effective in reducing the communication overhead in service discovery. However, since overlay network is independent of physical network topology and existing topology-aware mechanisms are based on the wired network, they are inefficient in MANET. We propose a proximity-based overlay network routing to overcome the inefficiency of routing in overlay network. In the proximity-based overlay network routing, each node collects information of physically close nodes by using one hop broadcast and routes messages to the logically closest node to destination. In a detailed ns-2 simulation study, we show that the proximity-based overlay network routing reduces the number of physical hops comparable to the flooding-based mechanism with low communication overhead. We also find that the proposed scheme works well in the mobile environment.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.7
• /
• pp.1923-1932
• /
• 1998

This paper deals with a 2-D objects recognition algorithm. And in this paper, we present an algorithm which can reduce the computation time in model retrieval by means of hashing technique instead of using the binary~tree method. In this paper, we treat an object boundary as a string of structural units and use an attributed string matching algorithm to compute similarity measure between two strings. We select from the privileged strings a privileged string wIth mmimal eccentricity. This privileged string is treated as the reference string. And thell we wllstructed hash table using the distance between privileged string and the reference string as a key value. Once the database of all model strings is built, the recognition proceeds by segmenting the scene into a polygonal approximation. The distance between privileged string extracted from the scene and the reference string is used for model hypothesis rerieval from the table. As a result of the computer simulation, the proposed method can recognize objects only computing, the distance 2-3tiems, while previous method should compute the distance 8-10 times for model retrieval.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.46 no.11
• /
• pp.1-11
• /
• 2009

A content based pub/sub (Publish/subscribe) services at the peer-to-peer network has the requirements about how to distribute contents information of subscriber and to delivery the events efficiently. For satisfying the requirements, a DHT(Distributed Hash Table) based pub/sub overlay networking and tree type topology based network construction using filter technique have been proposed. The DHT based technique is suitable for topic based pub/sub service but it's not good contents based service that has the variable requirements. And also filter based tree topology networking is not efficient at the environment where the user requirements are distributed. In this paper we propose the R-Tree algorithm based pub/sub overlay network construction method. The proposed scheme provides cost effective event delivery method by mapping user requirement to multi-dimension and hierarchical grouping of the requirements. It is verified by simulation at the variable environment of user requirements and events.

• Journal of Korea Multimedia Society
• /
• v.9 no.9
• /
• pp.1118-1130
• /
• 2006

Existed pattern analysis algorithms in data streams environment have researched performance improvement and effective memory usage. But when new data streams come, existed pattern analysis algorithms have to analyze patterns again and have to generate pattern tree again. This approach needs many calculations in real situation that needs real time pattern analysis. This paper proposes a method that continuously analyzes patterns of incoming data streams in real time. This method analyzes patterns fast, and thereafter obtains real time patterns by updating previously analyzed patterns. The incoming data streams are divided into several sequences based on time based window. Informations of the sequences are inputted into a hash table. When the number of the sequences are over predefined bound, patterns are analyzed from the hash table. The patterns form a pattern tree, and later created new patterns update the pattern tree. In this way, real time patterns are always maintained in the pattern tree. During pattern analysis, suffixes of both new pattern and existed pattern in the tree can be same. Then a pointer is created from the new pattern to the existed pattern. This method reduce calculation time during duplicated pattern analysis. And old patterns in the tree are deleted easily by FIFO method. The advantage of our algorithm is proved by performance comparison with existed method, MILE, in a condition that pattern is changed continuously. And we look around performance variation by changing several variable in the algorithm.