Catálogo de publicaciones - libros

This paper proposes novel fault models for ternary content addressable memories (TCAMs) regarding both physical defects and functional fault models. Novel test algorithms which guarantee high fault coverage and small test length for detecting the faults in high density TCAM cell array is also proposed. The proposed T algorithm is suitable for detecting physical faults with compact test patterns and provides 92% fault coverage with small test length. Also, the proposed T algorithm makes it possible to detect various bridging faults among adjacent cells of high density CAMs. The test access time is (10 * * 3 + 3 ) * write time and (12 * * 3) * compare time in a word by bit TCAM.

It is important to simulate a groundwater transport process, e.g., pollutant migration, through the vadose zone and subsequent mixing within the saturated zone to assess potential impacts of contaminants in the subsurface in preliminary stages. It is challenging to simulate heterogeneous soil characteristics and non-uniform initial contaminant concentration. This paper proposes a vertically heterogeneous model of combining heterogeneous vadose zone transport model and saturated zone mixing model using one-dimensional finite difference method to simulate the transport processes of liquid-phase advection, liquid- and vapor-phase dispersion, sorption, and decay of the contaminant. The proposed model was verified by comparing to an analytical solution and laboratory soil column experiments.

In this paper, we present a frame synchronization mechanism for a parallel rendering system to create an immersive virtual environment like CAVE. Our goal is to create a synchronized multi-display parallel rendering system built by the PC cluster. We implemented the proposed frame synchronization algorithm in both hardware and software approaches to investigate whether a special hardware is needed for the PC-CAVE architecture. Our experiments show that the hardware approach can achieve more accurate frame synchronization effect but the software approach provides better flexibility in building such a system.

This paper investigates system attributes and the corresponding performances of three Web cluster architectures, which have different traffic managers: the Alteon Web switch, the ALBM Traffic Manager, and a Linux Virtual Server. The Alteon Web switch is a hardware solution and the others provide software solutions to traffic managers that control incoming traffic in web clusters. In order to compare performances of these Web cluster architectures, we perform a set of experiments with various load scheduling algorithms by using a number of classified workloads. The experimental results compare the peak performances of the Web clusters and show the effects of workload types and load scheduling algorithms on performance trends.

RPR (Resilient Packet Ring) is a protocol proposed for the efficient delivery of frames over the metro network and is required to provide the bridging functionality for the interconnection of other external LANs. To do this, the node with the RPR capability should broadcast frames received from external networks to the entire RPR network. Since the RPR is proposed to operate in a metro network, most of RPR nodes are not hosts but bridges. Therefore broadcasting of all frames from external networks may significantly deteriorate the performance of the RPR network. In this paper, we propose a mechanism in which an RPR bridge node can efficiently support the bridging functionality by using a cache table with the information on external nodes.

Network security simulator is required for the study on the cyber intrusion and defense as cyber terrors have been increasingly popular. Until now, network security simulation aims at estimation of a small-size network or performance analysis of information protection systems. However, a systematic way to develop network security simulation is needed. In this paper, an ordered UML-based scenario model and its network security simulator are proposed. A network security scenario is presented as a case study of our proposed simulation model.

In this paper, we present a 3D tooth visualization system for the simulation of endodontic, orthodontic, and other dental treatments. Detail 3D model of individual tooth in mandible can be simulated and it can be rotated virtually around any axis to determine the best approach of surgery. To obtain higher accuracy of the reconstructed 3D model, a new segmentation method based on adaptive optimal thresholding and genetic snake is applied on the 2D CT images to separate individual tooth region as well as to extract the tooth contour. Surface reconstruction phase utilizes the contour data and tile the vertices of the contours of successive slices of CT images to form the triangular surface mesh of tooth. The paper, therefore, reports on experimentation with novel interactive techniques designed to work with the available geometric representation to teeth.

In this paper we analyze facial configuration for specifying young Korean men’s 3D virtual face. The approach is performed by defining facial feature points based on muscle-based animation, sampling of the images taken, conducting alignment of the feature vertices on the reference mesh model to feature points on the sampling images and clustering on the feature vertices. Finally we analyze comparative metric, which draws the reasonable conclusion that each mesh model represents highly distinctive patterns.