• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.6
• /
• pp.850-858
• /
• 2022

To deploy Gate Recurrent Units (GRU) on resource-constrained embedded devices, this paper presents a reconfigurable FPGA-based GRU accelerator that enables structured compression. Firstly, a dense GRU model is significantly reduced in size by hybrid quantization and structured top-k pruning. Secondly, the energy consumption on external memory access is greatly reduced by the proposed reuse computing pattern. Finally, the accelerator can handle a structured sparse model that benefits from the algorithm-hardware co-design workflows. Moreover, inference tasks can be flexibly performed using all functional dimensions, sequence length, and number of layers. Implemented on the Intel DE1-SoC FPGA, the proposed accelerator achieves 45.01 GOPs in a structured sparse GRU network without batching. Compared to the implementation of CPU and GPU, low-cost FPGA accelerator achieves 57 and 30x improvements in latency, 300 and 23.44x improvements in energy efficiency, respectively. Thus, the proposed accelerator is utilized as an early study of real-time embedded applications, demonstrating the potential for further development in the future.

• Journal of IKEEE
• /
• v.25 no.4
• /
• pp.664-671
• /
• 2021

This paper proposes a GPGPU structure that can reduce the number of operations and memory access by effectively applying a data reuse method to a convolutional neural network(CNN). Convolution is a two-dimensional operation using kernel and input data, and the operation is performed by sliding the kernel. In this case, a reuse method using an internal register is proposed instead of loading kernel from a cache memory until the convolution operation is completed. The serial operation method was applied to the convolution to increase the effect of data reuse by using the principle of GPGPU in which instructions are executed by the SIMT method. In this paper, for register-based data reuse, the kernel was fixed at 4×4 and GPGPU was designed considering the warp size and register bank to effectively support it. To verify the performance of the designed GPGPU on the CNN, we implemented it as an FPGA and then ran LeNet and measured the performance on AlexNet by comparison using TensorFlow. As a result of the measurement, 1-iteration learning speed based on AlexNet is 0.468sec and the inference speed is 0.135sec.

• Journal of Korea Society of Industrial Information Systems
• /
• v.15 no.2
• /
• pp.31-39
• /
• 2010

VoD(Video-on-Demand) servers have to provide timely processing guarantees for continuous media and reduce the storage and bandwidth requirements for continuous media. The compression techniques make the bit rates of compressed video data significantly variable from frame to frame. A VoD system should be able to provide the client with interactive operations such as fast forward and fast rewind in addition to normal playback of movie. However, interactive operations require additional resources such as storage space, disk bandwidth, memory and network bandwidth. In a stored video application such as VoD system, it is possible that a priori disk access patterns can be used to reserve the system resources in advance. In addition, clients of VoD server spend most of their time in playback mode and the period of time spent in interactive mode is relatively small. In this paper, I present the new buffer management scheme that provides efficient support for interactive operations in a VoD server using variable bit rate continuous media. Simulation results show that our strategy achieves 34% increase of the number of accepted clients over the LRU strategy.

• Journal of People, Plants, and Environment
• /
• v.22 no.5
• /
• pp.455-466
• /
• 2019

This study was carried out to analyze trends in floral products and services through the big data analysis of various social networking services (SNSs) and then to provide objective marketing directions for the floricultural industry. To analyze the big data of SNSs, we used four analytical methods: Cotton Trend (Social Matrix), Naver Big Data Lab, Instagram Big Data Analysis, and YouTube Big Data Analysis. The results of the big data analysis showed that SNS users paid positive attention to flower one-day classes that can satisfy their needs for direct experiences. Consumers of floral products and services had their favorite designs in mind and purchased floral products very actively. The demand for flower items such as bouquets, wreaths, flower baskets, large bouquets, orchids, flower boxes, wedding bouquets, and potted plants was very high, and cut flowers such as roses, tulips, and freesia were most popular as of June 1, 2019. By gender of consumers, females (68%) purchased more flower products through SNSs than males (32%). Consumers preferred mobile devices (90%) for online access compared to personal computers (PCs; 10%) and frequently searched flower-related words from February to May for the past three years from 2016 to 2018. In the aspect of design, they preferred natural style to formal style. In conclusion, future marketing activities in the floricultural industry need to be focused on social networks based on the results of big data analysis of popular SNSs. Florists need to provide consumers with the floricultural products and services that meet the trends and to blend them with their own sensitivity. It is also needed to select SNS media suitable for each gender and age group and to apply effective marketing methods to each target.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.87-109
• /
• 2020

Currently, thanks to the major stride made in developing wired and wireless communication technology, a variety of IT services are available on land. This trend is leading to an increasing demand for IT services to vessels on the water as well. And it is expected that the request for various IT services such as two-way digital data transmission, Web, APP, etc. is on the rise to the extent that they are available on land. However, while a high-speed information communication network is easily accessible on land because it is based upon a fixed infrastructure like an AP and a base station, it is not the case on the water. As a result, a radio communication network-based voice communication service is usually used at sea. To solve this problem, an additional frequency for digital data exchange was allocated, and a ship ad-hoc network (SANET) was proposed that can be utilized by using this frequency. Instead of satellite communication that costs a lot in installation and usage, SANET was developed to provide various IT services to ships based on IP in the sea. Connectivity between land base stations and ships is important in the SANET. To have this connection, a ship must be a member of the network with its IP address assigned. This paper proposes a SANET-CC protocol that allows ships to be assigned their own IP address. SANET-CC propagates several non-overlapping IP addresses through the entire network from land base stations to ships in the form of the tree. Ships allocate their own IP addresses through the exchange of simple requests and response messages with land base stations or M-ships that can allocate IP addresses. Therefore, SANET-CC can eliminate the IP collision prevention (Duplicate Address Detection) process and the process of network separation or integration caused by the movement of the ship. Various simulations were performed to verify the applicability of this protocol to SANET. The outcome of such simulations shows us the following. First, using SANET-CC, about 91% of the ships in the network were able to receive IP addresses under any circumstances. It is 6% higher than the existing studies. And it suggests that if variables are adjusted to each port's environment, it may show further improved results. Second, this work shows us that it takes all vessels an average of 10 seconds to receive IP addresses regardless of conditions. It represents a 50% decrease in time compared to the average of 20 seconds in the previous study. Also Besides, taking it into account that when existing studies were on 50 to 200 vessels, this study on 100 to 400 vessels, the efficiency can be much higher. Third, existing studies have not been able to derive optimal values according to variables. This is because it does not have a consistent pattern depending on the variable. This means that optimal variables values cannot be set for each port under diverse environments. This paper, however, shows us that the result values from the variables exhibit a consistent pattern. This is significant in that it can be applied to each port by adjusting the variable values. It was also confirmed that regardless of the number of ships, the IP allocation ratio was the most efficient at about 96 percent if the waiting time after the IP request was 75ms, and that the tree structure could maintain a stable network configuration when the number of IPs was over 30000. Fourth, this study can be used to design a network for supporting intelligent maritime control systems and services offshore, instead of satellite communication. And if LTE-M is set up, it is possible to use it for various intelligent services.

• Journal of KIISE:Computing Practices and Letters
• /
• v.12 no.2
• /
• pp.91-101
• /
• 2006

Researches on services of vehicles have been mainly focused on how to provide useful information and entertainment for an in-vehicle driver. However, the needs are appreciably increased for more advanced services that help drivers to check and manage their vehicles conveniently, without requiring drivers to attach to their vehicles. It is a sort of ubiquitous computing, providing an intelligent interactive services for human at any time and any where. In this paper, we present an intelligent vehicle information system to enable a driver to remotely diagnose and control a vehicle via CDMA communication network connected to the Internet. The system improves mobility for diagnosis and control of vehicle by implementing a cut and call back mechanism, which allows the vehicle terminal to have access to the information server on the Internet via CDMA call. No matter where the driver is, he can obtain the remote diagnosis and control services on the web browser without any additional application installation. Design methodology is introduced and evaluation results are analyzed for the CDMA-based intelligent vehicle information system. The experimental results show that the response time of the vehicle terminal to a web client request is 10.302 seconds at the beginning and 646.44ms thereafter. The average response time of CAN sensor node to a vehicle terminal request is 6.669ms.

• Journal of KIISE:Computing Practices and Letters
• /
• v.10 no.3
• /
• pp.221-230
• /
• 2004

In general, remote monitoring systems monitor the status of distributed hosts and/or applications in real-time for diverse managerial purposes. However, most of the extant systems have a few undesirable problems. First of all, they are platform-dependent and are not resilient to network and/or host failures. Moreover, they normally focus on the resource usage trends in monitored hosts, rather than on the status change of the applications running on them. We strongly believe that the latter has more direct and profound effect on the resource usage patterns on each host. In this paper, we present the design and implementation of the Remote Applications Monitoring System (RAMS) that enables us to effectively manage distributed applications through a real-time monitoring of their respective resource usages. The RAMS is a centralized system that consists of many distributed agents and a single centralized manager. An agent on each host is in charge of collecting and reporting the status of local applications. The manager handles agent registration and provides a central access point to the selection and monitoring of distributed applications. The salient features of the system include robustness and portability The adoption of JINI greatly facilitates an automatic recovery from partial network failure and host failure.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.9 no.4
• /
• pp.824-831
• /
• 2005

Past corporaion's network security system is single security solution, or mixed several ways, but there was inefficient system because doing not get into organic link But, constructed more strong security system by ESM enterance on. ESM uses way to integrate of each agent to manage easily various kinds security solution. That is, it is system that connect system of existent VPN, FireWall, IDS and so on configurationally depending on security policy and manage. ESM is security system that is developed more than existent security system. But, practical use of network and the development speed of technology being increasing with the mon faster speed, is heightening the level more as well as dysfunction of information crime and so on. Many improvements are required at ESM system, this research wished to make up for the weak-point in the ESM system about interior security. Studied on structure of security solution that is basis of security policy. VPN, Firewall, IDS's link that is main composition of existing security system analysis, reconstructed. And supplemented security of ESM system itself. Establish imaginary intrusion and comparative analysis access data that apply each Telnet Log analysys IDS existent ESM system and proposed ESM system comparative analysis. Confirm the importance of interior security and inspected security of proposed system.

• Spatial Information Research
• /
• v.5 no.2
• /
• pp.147-160
• /
• 1997

"The "Map Algebra", beeing recognized as a viable theoretical framework for GIS (Geographica Infonnation System), models map layers as "operands" which are the basic unit of geo-processing, and a variety of GIS commands as "operators." In this paper, we attempt at lifting some limitations of map algebras proposed in GIS literature. First, we model map layer as "function" such that we may employ the notion of meta operator (or, higher-order funtion) available in the functional programming paradigm. This approach provides map algebraic language with "programmability" needed in GIS user language. Second, we extend the semantics of, and improve on the sytactic structure of map algebraic language. Mer the data model and language associated with map algebra are formalized, we proceed to design and implement a prototype of map algebraic processor. The parser of the language in our prototype plays the role of transforming the native and heterogeneous user language of current GISs into a canonical map algebraic language. The prototype, named "MapSee" is a proof-of-concept system for the ideas we propsed in this paper. We believe that the uniform interface based on the map algebraic language will make promising infrastructure to support "Internet GIS." This is because the uniform but powerful interface through the Web clients allow access to both geo-data and geo-processing resources distributed over the network.to both geo-data and geo-processing resources distributed over the network.

• The KIPS Transactions:PartC
• /
• v.16C no.2
• /
• pp.151-164
• /
• 2009

The importance of sensor networks as a base of ubiquitous computing realization is being highlighted, and espicially the security is recognized as an important research isuue, because of their characteristics.Several efforts are underway to provide security services in sensor networks, but most of them are preventive approaches based on cryptography. However, sensor nodes are extremely vulnerable to capture or key compromise. To ensure the security of the network, it is critical to develop security Intrusion Detection System (IDS) that can survive malicious attacks from "insiders" who have access to keying materials or the full control of some nodes, taking their charateristics into consideration. In this perper, we design a distributed and adaptive IDS architecture on sensor networks, respecting both of energy efficiency and IDS efficiency. Utilizing a modified HEED algorithm, a clustering algorithm, distributed IDS nodes (dIDS) are selected according to node's residual energy and degree. Then the monitoring results of dIDSswith detection codes are transferred to dIDSs in next round, in order to perform consecutive and integrated IDS process and urgent report are sent through high priority messages. With the simulation we show that the superiorities of our architecture in the the efficiency, overhead, and detection capability view, in comparison with a recent existent research, adaptive IDS.