• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.7
• /
• pp.2977-2997
• /
• 2018

In big data era, fresh data grows rapidly every day. More than 30,000 gigabytes of data are created every second and the rate is accelerating. Many organizations rely heavily on real time streaming, while big data stream computing helps them spot opportunities and risks from real time big data. Storm, one of the most common online stream computing platforms, has been used for big data stream computing, with response time ranging from milliseconds to sub-seconds. The performance of Storm plays a crucial role in different application scenarios, however, few studies were conducted to evaluate the performance of Storm. In this paper, we investigate the performance of Storm under different application scenarios. Our experimental results show that throughput and latency of Storm are greatly affected by the number of instances of each vertex in task topology, and the number of available resources in data center. The fault-tolerant mechanism of Storm works well in most big data stream computing environments. As a result, it is suggested that a dynamic topology, an elastic scheduling framework, and a memory based fault-tolerant mechanism are necessary for providing high throughput and low latency services on Storm platform.

• Journal of Information Technology Services
• /
• v.3 no.1
• /
• pp.101-115
• /
• 2004

The phrase "Ubiquitous Computing" has become popular ever since Mark Weiser used it in an article. It is to realize a computerized environment in which small computers are embedded and cooperate with each other. This environment will support many activities of our daily life. In a Ubiquitous Computing environment, various devices will be connected to the network from houses and buildings. Therefore it is necessary to ensure network security and to protect private data from tapping, falsification and the disguising of identity by others. This study reviews the Ubiquitous Computing technologies in detail and outlines the design concept of the Stream Cipher Algorithm.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2000.04a
• /
• pp.1166-1171
• /
• 2000

기존의 멀티미디어 연구의 실현에 있어 가장 큰 문제라 할 수 있던 성능의 문제가 하드웨어의 급속한 발달로 해결되어 감에 따라 멀티미디어 및 제반 관련기술도 함께 발전되었으며 이에 기반한 multimedia stream에서의 event를 검출하기 위한 다양한 연구들이 진행되어 왔다. 그러나 지금까지의 연구는 주로 전송 및 저장, 검색에 집중되어 연구되어 왔으며 영상인식 등의 Vision관련 연구에서는 멀티미디어 스트리밍 기술과의 연동을 고려하지 않은 연구를 수행함에 따라 검출 가능한 event가 있다고 하더라도 응용영역에 종속적인 인테페이스만을 고려함에 따라 사용자가 이를 기술(記述, description)하거나, 사용자에게 검출 가능한 event를 제시하기 위해 일반화된 방법이 제시되어 있지 않았다. 본 연구에서는 사용자가 검출을 원하는 event를 기술하는 방법과, 시스템에서 검출 가능한 event를 제시하기 위한 방법을 제안하고, 제시되는 방법이 응용영역에 독립적이기 위해 요구되는 사항들과 객체 단위인 이벤트/행위와 처리기 사이의 인터페이스에 관하여 정의한 후 기본적인 동작방식을 제안한다.

• Journal of Korea Spatial Information System Society
• /
• v.11 no.1
• /
• pp.9-16
• /
• 2009

In ubiquitous spatial computing environments, GeoSensor generates sensor data streams including spatial information as well as various conventional sensor data from RFID, WSN, Web CAM, Digital Camera, CCTV, and Telematics units. This GeoSensor enables the revitalization of various ubiquitous USN technologies and services on geographic information. In order to service the u-GIS applications based on GeoSensors, it is indispensable to efficiently process sensor data streams from GeoSensors of a wide area. In this paper, we propose a GeoSensor data stream processing system for u-GIS computing over real-time stream data from GeoSensors with geographic information. The proposed system provides efficient gathering, storing, and continuous query processing of GeoSensor data stream, and also makes it possible to develop diverse u-GIS applications meet each user requirements effectively.

• 한국정보통신설비학회:학술대회논문집
• /
• 2004.08a
• /
• pp.137-141
• /
• 2004

네트워크 사용자의 급속한 증가로 네트워크 내의 부하를 감당하기에는 많은 어려움을 가져왔다. 이와 같은 이유로 기존의 TCP/IP에서 세션을 통하여 노드들 간의 통신을 연결하는 방식에서 현재는 하나의 채널을 통해 고속의 I/O가 가능하도록 하는 기술이 많이 연구되고 있다. 그 대표적인 것으로 인피니밴드가 있다. 인피니밴드는 프로세싱 노드와 입출력 장치 사이의 통신, 프로세스간 통신에 대한 산업 표준이 되고 있고 프로세싱 노드와 입출력 장치를 연결하기 위해 스위치 기반의 상호 연결은 전통적인 버스 입출력을 대체하는 새로운 입출력 방식이 사용된다. 또한 인피니밴드에서는 현재 많은 이슈가 되고 있는 RDMA 방식을 이용해 원격지 서버들 간에 직접 메모리 접근 방식을 통해 CPU와 OS의 로드를 최소화하고 있다. 본 논문에서는 RDMA를 적용한 새로운 채널 기반 네트웍의 프로토콜인 SDP(Socket Direct Protocol)를 구현하여 SDP_STREAM의 패킷 처리량에 대한 성능을 평가한다. 그리고 이에 대한 성능 평가를 위해서 Netperf 툴을 이용했다. 특히 Zero-Copy방식을 사용하지 않는 일반적인 소켓 API을 이용한 TCP_STREAM과 Zero-Copy방식을 이용한 SDP_STREAM의 패킷 처리량을 비교했으며 성능 평가 결과는 기존의 TCP_STREAM 패킷 처리량에 비해 약 3배 이상 향상된 결과를 나타냈다.

• Journal of Korean Society on Water Environment
• /
• v.31 no.5
• /
• pp.491-506
• /
• 2015

Distributed models represent watersheds using a network of numerous, uniform calculation units to provide spatially detailed and consistent evaluations across the watershed. However, these models have a disadvantage in general requiring a high computing cost. Semi-distributed models, on the other hand, delineate watersheds using a simplified network of non-uniform calculation units requiring a much lower computing cost than distributed models. Employing a simplified network of non-uniform units, however, semi-distributed models cannot but have limitations in spatially-consistent simulations of hydrogeochemical processes and are often not favoured for such a task as identifying critical source areas within a watershed. Aiming to overcome these shortcomings of both groups of models, a hybrid watershed model STREAM (Spatio-Temporal River-basin Ecohydrology Analysis Model) was developed in this study. Like a distributed model, STREAM divides a watershed into square grid cells of a same size each of which may have a different set of hydrogeochemical parameters reflecting the spatial heterogeneity. Like many semi-distributed models, STREAM groups individual cells of similar hydrogeochemical properties into representative cells for which real computations of the model are carried out. With this hybrid structure, STREAM requires a relatively small computational cost although it still keeps the critical advantage of distributed models.

• Journal of Internet Computing and Services
• /
• v.18 no.3
• /
• pp.1-9
• /
• 2017

Smart Cities intelligently manage numerous infrastructures, including Smart-City IoT devices, and provide a variety of smart-city applications to citizen. In order to provide various information needed for smart-city applications, Smart Cities require a function to intelligently process large-scale streamed big data that are constantly generated from a large number of IoT devices. To provide smart services in Smart-City, the Smart-City Consortium uses stream reasoning. Our stream reasoning requires real-time processing of big data. However, there are limitations associated with real-time processing of large-scale streamed big data in Smart Cities. In this paper, we introduce one of our researches on cloud computing based real-time distributed-parallel-processing to be used in stream-reasoning of IoT big data in Smart Cities. The Smart-City Consortium introduced its previously developed smart-city middleware. In the research for this paper, we made cloud computing based real-time distributed-parallel-processing available in the cloud computing platform of the smart-city middleware developed in the previous research, so that we can perform real-time distributed-parallel-processing with them. This paper introduces a real-time distributed-parallel-processing method and system for stream reasoning with IoT big data transmitted from various sensors of Smart Cities and evaluate the performance of real-time distributed-parallel-processing of the system where the method is implemented.

• Journal of Korea Spatial Information System Society
• /
• v.9 no.1
• /
• pp.91-104
• /
• 2007

In ubiquitous computing environment, people expect services to change actively in response to the change of themselves and surrounding environment. In order to meet such expectations, the streaming service needs technologies for migrating through various devices, following the user's movement, and the technology of individualized streaming push service. In this paper, we designed and implemented U-Stream, a ubiquitous streaming service system, based on user-defined context and stream view. The system classifies contexts according to streaming service migration and allows users to define some contexts and, by doing so, provides user-oriented individualized services rather than uniform services based solely on the developers' anticipations.

• Journal of Computing Science and Engineering
• /
• v.5 no.1
• /
• pp.85-110
• /
• 2011

Achieving situation awareness is especially challenging for real-time data stream applications because they i) operate on continuous unbounded streams of data, and ii) have inherent realtime requirements. In this paper we showed how formal data stream modeling and analysis can be used to better understand stream behavior, evaluate query costs, and improve application performance. We used MEDAL, a formal specification language based on Petri nets, to model the data stream queries and the quality-of-service management mechanisms of RT-STREAM, a prototype system for data stream management. MEDAL's ability to combine query logic and data admission control in one model allows us to design a single comprehensive model of the system. This model can be used to perform a large set of analyses to help improve the application's performance and quality of service.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.878-879
• /
• 2021

Most state-of-the-art CNNs for action recognition are based on a two-stream architecture: RGB frames stream represents the appearance and the optical flow stream interprets the motion of action. However, the cost of optical flow computation is very high and then it increases action recognition latency. We introduce a design strategy for action recognition inspired by a two-stream network and teacher-student architecture. There are two sub-networks in our neural networks, the optical flow sub-network as a teacher and the RGB frames sub-network as a student. In the training stage, we distill the feature from the teacher as a baseline to train student sub-network. In the test stage, we only use the student so that the latency reduces without computing optical flow. Our experiments show that its advantages over two-stream architecture in both speed and performance.