• 한국정보통신학회논문지
• /
• 제25권1호
• /
• pp.50-55
• /
• 2021

다중경로 시스템은 유선망, LTE망, 위성망 등 다양한 망을 동시에 활용하여 데이터를 전송하는 시스템으로, 통신망의 전송속도, 신뢰도, 보안성 등을 높이기 위해 제안되었다. 본 논문에서는 이 시스템에서 각 망의 지연시간을 보상으로 하는 강화학습 기반 경로 선택 방안을 제안하고자 한다. 기존의 강화학습 모델과는 다르게, deep Q 학습을 이용하여 망의 변화하는 환경에 즉각적으로 대응하도록 알고리즘을 설계하였다. 네트워크 환경에서는 보상 정보를 일정 지연시간이 지나야 얻을 수 있으므로 이를 보정하는 방안 또한 함께 제안하였다. 성능을 평가하기 위해, 분산 데이터베이스와 텐서플로우 모듈 등을 포함한 테스트베드 학습 서버를 개발하였다. 시뮬레이션 결과, 제안 알고리즘이 RTT 감소 측면에서 최저 지연시간을 선택하는 방안보다 20% 가량 좋은 성능을 가지는 것을 확인하였다.

• 한국정보처리학회논문지
• /
• 제7권3호
• /
• pp.799-807
• /
• 2000

In this paper, object replication Client/server under distributed computing system is design and implementation. Today many end-users have a computer communication by using internet in the distributed system of client/server. If many users request services to a specific remote server, the server should have got a overhead for hat service processing, delayed the speed for replay, and bring a bottleneck in communication network. Therefore object replication method was proposed to solve this problems. The growth of internet works and distributed applications has increased the need for large scale replicated systems. However, existing replication protocols do not address scale and autonomy issues adequately. Further, current application protocol require consistency of different levels, and therefore should be the selection function of consistency in them, in order to have particular semantics of each level. In this paper, server overhead and bottleneck happening in remote procedure call be using server object replication. Therefore access transparency can be improved by sharing object duplicately. So it will Keep up with the consistency within the replicated objects.

• 대한임베디드공학회논문지
• /
• 제17권5호
• /
• pp.297-307
• /
• 2022

The simple procedural segment selection algorithm commonly used in Dynamic Adaptive Streaming over HTTP (DASH) reveals severe weakness to provide high-quality streaming services in the integrated mobile networks of various wired and wireless links. A major issue could be how to properly cope with dynamically changing underlying network conditions. The key to meet it should be to make the segment selection algorithm much more adaptive to fluctuation of network traffics. This paper presents a system architecture that replaces the existing procedural segment selection algorithm with a deep reinforcement learning algorithm based on the Asynchronous Advantage Actor-Critic (A3C). The distributed A3C-based deep learning server is designed and implemented to allow multiple clients in different network conditions to stream videos simultaneously, collect learning data quickly, and learn asynchronously, resulting in greatly improved learning speed as the number of video clients increases. The performance analysis shows that the proposed algorithm outperforms both the conventional DASH algorithm and the Deep Q-Network algorithm in terms of the user's quality of experience and the speed of deep learning.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제7권1호
• /
• pp.22-37
• /
• 2013

Content distribution to a large number of concurrent clients stresses both server and network. While the server limitation can be circumvented by deploying server clusters, the network limitation is far less easy to cope with, due to the difficulty in measuring and balancing network load. In this paper, we use two useful network load metrics, the worst link stress (WLS) and the degree of interference (DOI), and formulate the problem as partitioning the clients into disjoint subsets subject to the server capacity constraint so that the WLS and the DOI are reduced for each session and also well balanced across the sessions. We present a network load-aware partition algorithm, which is practicable and effective in achieving the design goals. Through experiments on PlanetLab, we show that the proposed scheme has the remarkable advantages over existing schemes in reducing and balancing the network load. We expect the algorithm and performance metrics can be easily applied to various Internet applications, such as media streaming, multicast group member selection.

• 한국인터넷방송통신학회논문지
• /
• 제22권6호
• /
• pp.107-112
• /
• 2022

드론과 같은 임베디드 시스템에서 데이터를 서버로 전송해 실시간 분석을 진행함에 있어진행하는 데 초분광 영상 전체를 저장, 전송, 분석하는 데 전력 소모와 시간이 많이 소요되어 어려움이 있다. 그래서 초분광 영상 데이터는 차원 축소 또는 압축 전처리를 통해 서버로 전송하게 된다. 분석에 필요한 밴드만 보내기 위해서는 피처 선택 기법을 사용하는데 이러한 알고리즘은 대게 효율은 높더라도 영상 크기에 따라 처리 시간이 매우 소요가 크다. 본 논문에서는 밴드선택 알고리즘의 시간적인 단점을 개선하여한 기댓값 기반의 알고리즘을 제안한다. 실험 결과 8GB 데이터의 40000*682 해상도 이미지 기준 평균 소요 시간인 24시간을 60~180초 내외로 감소시키고, 150개 밴드 중에 45개를 활용하여 7.6GB 램 사용을 2.3GB로 크게 감소시켰다. 시간은 크게 줄였음에도 픽셀 분류 성능은 기존과 유사하게 98% 이상의 분석 결과를 도출하였다.

• International Journal of Advanced Culture Technology
• /
• 제9권1호
• /
• pp.218-223
• /
• 2021

Pervasive enhancement and required enforcement of the Internet of Things (IoTs) in a distributed massively multiplayer online architecture have effected in massive growth of Big-Data in terms of server over-load. There have been some previous works to overcome the overloading of server works. However, there are lack of considered methods, which is commonly applicable. Therefore, we propose a combing Sparse Auto-Encoder and Load-Balancing, which is ProGReGA for Big-Data of server loads. In the process of Sparse Auto-Encoder, when it comes to selection of the feature-pattern, the less relevant feature-pattern could be eliminated from Big-Data. In relation to Load-Balancing, the alleviated degradation of ProGReGA can take advantage of the less redundant feature-pattern. That means the most relevant of Big-Data representation can work. In the performance evaluation, we can find that the proposed method have become more approachable and stable.

• 정보처리학회논문지C
• /
• 제10C권1호
• /
• pp.33-38
• /
• 2003

최근 들어 다수의 클라이언트에게 컨텐츠를 안정적으로 서비스하기 위하여 여러 부본서버를 운영한다. 이 부본서버 중 시공간적으로 다양한 환경으로부터 요청된 클라이언트의 요구를 가장 효과적으로 처리할 수 있는 서버를 선택하기 위한 서버 선택 알고리즘이 여러 분야에서 활발히 연구되고 있다. 본 논문은 비교적 큰 용량의 컨텐츠를 서비스하는 부본서버 간의 하드웨어 성능 차이가 큰 경우에 QoS를 향상시킬 수 있는 방법을 제안하였다. 또, 부본서버의 하드웨어 성능평가 알고리즘과 HTTP 응답시간의 평균과 편차를 동시에 고려한 알고리즘을 제안하였다. 그리고 시뮬레이션을 통하여 일반적인 경우에 가장 성능이 우수한 것으로 알려진 HTTP 반응시간 평균과 편차에 근거한 부본선택 알고리즘이 비교적 큰 사이즈의 컨텐츠 서비스할 경우와 부본 서버들간의 하드웨어 성능 차가 큰 경우에 더 나은 성능을 보인다는 것을 확인하였다.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2008년도 추계학술대회A
• /
• pp.795-800
• /
• 2008

The paper presents research for the established experiment environment of multi vehicle robot, localization algorithm that is based on vehicle control, and path tracking. The established experiment environment consists of ultrasonic positioning system, vehicle robot, server and wireless module. Ultrasonic positioning system measures positioning for using ultrasonic sensor and generates many errors because of the influence of environment such as a reflection of wall. For a solution of this fact, localization algorithm is proposed to determine a location using vehicle kinematics and selection of a reliable location data. And path tracking algorithm is proposed to apply localization algorithm and LOS, finally, that algorithms are verified via simulation and experimental

• 한국정보처리학회논문지
• /
• 제6권5호
• /
• pp.1189-1202
• /
• 1999

As compared with VOD data, NOD article data has the following characteristics: it is created at any time, has a short life cycle, is selected as not one article but several articles by a user, and has high access locality in time. Because of these intrinsic features, user access patterns of NOD article data are different from those of VOD. Thus, building NOD system using the existing techniques of VOD system leads to poor performance. In this paper, we analysis the log file of a currently running electronic newspaper, show that the popularity distribution of NOD articles is different from Zipf distribution of VOD data, and suggest a new popularity model of NOD article data MS-Zipf(Multi-Selection Zipf) distribution and its approximate solution. Also we present a life cycle model of NOD article data, which shows changes of popularity over time. Using this life cycle model, we develop LLBF (Largest Life-cycle Based Frequency) prefetching algorithm and analysis he performance by simulation. The developed LLBF algorithm supports the similar level in hit-ratio to the other prefetching algorithms such as LRU(Least Recently Used) etc, while decreasing the number of data replacement in article prefetching and reducing the overhead of the prefetching in system performance. Using the accurate user access patterns of NOD article data, we could analysis correctly the performance of NOD server system and develop the efficient policies in the implementation of NOD server system.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권6호
• /
• pp.2282-2303
• /
• 2021

To solve the problems of heavy computing load and system transmission pressure in energy internet (EI), we establish a three-tier cloud-edge integrated EI network based on a cloud-edge collaborative computing to achieve the tradeoff between energy consumption and the system delay. A joint optimization problem for resource allocation and task offloading in the threetier cloud-edge integrated EI network is formulated to minimize the total system cost under the constraints of the task scheduling binary variables of each sensor node, the maximum uplink transmit power of each sensor node, the limited computation capability of the sensor node and the maximum computation resource of each edge server, which is a Mixed Integer Non-linear Programming (MINLP) problem. To solve the problem, we propose a joint task offloading and resource allocation algorithm (JTOARA), which is decomposed into three subproblems including the uplink transmission power allocation sub-problem, the computation resource allocation sub-problem, and the offloading scheme selection subproblem. Then, the power allocation of each sensor node is achieved by bisection search algorithm, which has a fast convergence. While the computation resource allocation is derived by line optimization method and convex optimization theory. Finally, to achieve the optimal task offloading, we propose a cloud-edge collaborative computation offloading schemes based on game theory and prove the existence of Nash Equilibrium. The simulation results demonstrate that our proposed algorithm can improve output performance as comparing with the conventional algorithms, and its performance is close to the that of the enumerative algorithm.