• JSTS:Journal of Semiconductor Technology and Science
• /
• v.17 no.4
• /
• pp.505-513
• /
• 2017

As mobile systems are performing various functionality in the IoT (Internet of Things) era, network-on-chip (NoC) plays a pivotal role to support communication between the tens and in the future potentially hundreds of interacting modules in system-on-chips (SoCs). Owing to intensive research efforts more than a decade, NoCs are now widely adopted in various SoC designs. Especially, studies on application-specific NoCs (ASNoCs) that consider the heterogeneous nature of modern SoCs contribute a significant share to use of NoCs in actual SoCs, i.e., ASNoC connects non-uniform processing units, memory, and other intellectual properties (IPs) using flexible router positions and communication paths. Although it is not difficult to find the prior works on ASNoC synthesis and optimization, little research has addressed the issues how to convert different protocols and data widths to make a NoC compatible with various IPs. Thus, in this paper, we address important issues on ASNoC implementation to support and convert multiple interfaces. Based on the in-depth discussions, we finally introduce our FPGA-proven full-custom ASNoC.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.15 no.3
• /
• pp.119-127
• /
• 2020

The detection and recognition of abnormal driving becomes crucial for achieving safety in Intelligent Transportation Systems (ITS). This paper presents a feature extraction method based on spectral data to train a neural network model for driving behavior recognition. The proposed method uses a two stage signal processing approach to derive time-saving and efficient feature vectors. For the first stage, the feature vector set is obtained by calculating variances from each frequency bin containing the power spectrum data. The feature set is further reduced in the second stage where an intersection method is used to select more significant features that are finally applied for training a neural network model. A stream of live signals are fed to the trained model which recognizes the abnormal driving behaviors. The driving behaviors considered in this study are weaving, sudden braking and normal driving. The effectiveness of the proposed method is demonstrated by comparing with existing methods, which are Particle Swarm Optimization (PSO) and Convolution Neural Network (CNN). The experiments show that the proposed approach achieves satisfactory results with less computational complexity.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.4B
• /
• pp.455-463
• /
• 2004

In this paper, we deal with a ring-mesh network design problem arising from the deployment of WDM for the optical internet. The ring-mesh network consists of ring topology and full mesh topology for satisfying traffic demand while minimizing the cost of OAOMs and OXCs. The problem seeks to find an optimal clustering of traffic demands in the network such that the total number of node assignments is minimized, while satisfying ring capacity and node cardinality constraints. We formulate the problem as a mixed-integer programming model and prescribe a tabu search heuristic procedure Promising computational results within 3% optimality gap are obtained using the proposed method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.18 no.3
• /
• pp.339-347
• /
• 1993

The channel assignment problem in a mobile communication system is a NP-complete combinatorial optimization problem, in which the calculation time increases exponentially as the range of the problem is extended. This paper adapts a conventional Hopfield neural network model to the channel assignment problem to relieve the calculation time by means of the parallelism supplied from the neural network. In the simulation study, we checked the feasability of such a parallel method for the fixed channel assignment with uniform, and nouniform channel requirements, and for the dynamic channel assignment with considering continously varying channel requirements.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2004.11a
• /
• pp.272-275
• /
• 2004

플라즈마 식각공정은 소자제조를 위한 미세 패턴닝 제작에 이용되고 있다. 공정 메커니즘의 정성적 해석, 최적화, 그리고 제어를 위해서는 컴퓨터 예측모델의 개발이 요구된다. 역전파 신경망 (backpropagation neural network-BPNN) 모델을 개발하는 데에는 다수의 학습인자가 관여하고 있으며, 가장 그 최적화가 어려운 학습인자는 초기웨이트이다. 모델개발시, 초기웨이트는 random 값으로 설정이 되며, 이로 인해 초기웨이트의 최적화가 어렵다. 본 연구에서는 유전자 알고리즘 (genetic algorithm-GA)을 이용하여 BPNN의 초기웨이트를 최적화하였으며, 이를 식각공정 모델링에 적용하여 평가하였다. 실리카 식각공정 데이터는 $2^3$ 인자 실험계획법을 이용하여 수집하였으며, GA에 관여하는 두 확률인자의 영향을 42 인자 실험계획법을 이용하여 최적화 하였다. 종래의 모델에 비해, 최적화된 모델은 실리카 식각률, Al 식각률, Al 선택비, 그리고 프로파일 응답에 대해서 각 기 24%, 13%,, 16%, 그리고 17%의 향상률을 보였다. 이는 제안된 최적화 기법이 플라즈마 모델의 예측성능을 증진하는데 효과적으로 응용될 수 있음을 의미한다.

• Proceedings of the KIEE Conference
• /
• 2005.10b
• /
• pp.152-154
• /
• 2005

In this paper, we propose a new architecture of Fuzzy Polynomial Neural Networks(FPNN) by means of genetically optimized Fuzzy Polynomial Neuron(FPN) and discuss its comprehensive design methodology involving mechanisms of genetic optimization, especially Genetic Algorithms(GAs). The design of the network exploits the extended Group Method of Data Handling(GMDH) with some essential parameters of the network being provided by the designer and kept fixed throughout the overall development process. This restriction may hamper a possibility of producing an optimal architecture of the model. The proposed FPNN gives rise to a structurally optimized network and comes with a substantial level of flexibility in comparison to the one we encounter in conventional FPNNs. It is shown that the proposed genetic algorithms-based Fuzzy Polynomial Neural Networks is more useful and effective than the existing models for nonlinear process. We experimented with Medical Imaging System(MIS) dataset to evaluate the performance of the proposed model.

• ETRI Journal
• /
• v.39 no.3
• /
• pp.353-363
• /
• 2017

Recently, sink mobility has been shown to be highly beneficial in improving network lifetime in wireless sensor networks (WSNs). Numerous studies have exploited mobile sinks (MSs) to collect sensed data in order to improve energy efficiency and reduce WSN operational costs. However, there have been few studies on the effectiveness of MS operation on WSN closed operating cycles. Therefore, it is important to investigate how data is collected and how to plan the trajectory of the MS in order to gather data in time, reduce energy consumption, and improve WSN network lifetime. In this study, we combine two methods, the cluster-head election algorithm and the MS trajectory optimization algorithm, to propose the optimal MS movement strategy. This study aims to provide a closed operating cycle for WSNs, by which the energy consumption and running time of a WSN is minimized during the cluster election and data gathering periods. Furthermore, our flexible MS movement scenarios achieve both a long network lifetime and an optimal MS schedule. The simulation results demonstrate that our proposed algorithm achieves better performance than other well-known algorithms.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.12
• /
• pp.5759-5779
• /
• 2017

As the main means to cope with the stubborn problem of drunk driving, the inspection of drunk driving has already been paid more attention and thus reinforced. In this paper, we model this scenario as a Stackelberg game, where the police department (called defender) allocates resources dynamically in terms of the traffic situation on the traffic network to arrest drink drivers and drivers who drink (called attacker), whether choosing drunk driving or designated driving service, expect to minimize their cost for given travel routes. However, with the number of resources are limited, our goal is to calculate the optimal resource allocation strategy for the defender. Therefore, first, we provide an effective approach (named OISDD) to fulfill our goal, i.e., generate the optimal strategy to inspect drunk driving. Second, we apply OISDD to directed graphs (which are abstracted from Dalian traffic network) to analyze and test its correctness and rationality. The experimental results show that OISDD is feasible and efficient.

• Smart Structures and Systems
• /
• v.22 no.2
• /
• pp.175-183
• /
• 2018

Time-series data often contain one of the most valuable pieces of information in many fields including manufacturing. Because time-series data are relatively cheap to acquire, they (e.g., vibration signals) have become a crucial part of big data even in manufacturing shop floors. Recently, deep-learning models have shown state-of-art performance for analyzing big data because of their sophisticated structures and considerable computational power. Traditional models for a machinery-monitoring system have highly relied on features selected by human experts. In addition, the representational power of such models fails as the data distribution becomes complicated. On the other hand, deep-learning models automatically select highly abstracted features during the optimization process, and their representational power is better than that of traditional neural network models. However, the applicability of deep-learning models to the field of prognostics and health management (PHM) has not been well investigated yet. This study integrates the "residual fitting" mechanism inherently embedded in the wavelet transform into the convolutional neural network deep-learning structure. As a result, the architecture combines a signal smoother and classification procedures into a single model. Validation results from rotor vibration data demonstrate that our model outperforms all other off-the-shelf feature-based models.

• Journal of Institute of Control, Robotics and Systems
• /
• v.4 no.6
• /
• pp.802-810
• /
• 1998

The purpose of this study is to find the analytic solution of determining the optimal capacity of processes and storages to meet the product demand. Recent trend to reduce product delivery time and to provide high quality product to customer requires the increasing capacity of storage facilities. However, the cost of constructing and operating storage facilities is becoming substantial because of increasing land value, environmental and safety concern. Therefore, reasonable decision making about the capacity of processes and storages is important subject for industries. The industrial solution for this subject is to use the classical economic lot sizing method, EOQ(Economic Order Quantity) model, trimmed with practical experience but the unrealistic assumption of EOQ model is not suitable for the chemical plant design with highly interlinked processes and storages. This study, a first systematic attempt for this subject, clearly overcomes the limitation of classical lot sizing method. The superstructure of the plant consists of the network of serially and/or parallelly interlinked processes and storages. A novel production and inventory analysis method, PSW(Periodic Square Wave) model, is applied. The objective function of optimization is minimizing the total cost composed of setup and inventory holding cost. The advantage of PSW model comes from the fact that the model provide a set of simple analytic solution in spite of realistic description of material flow between process and storage. The resulting simple analytic solution can greatly enhance the proper and quick investment decision for the preliminary plant design confronting diverse economic situation.