Catálogo de publicaciones - libros

A course-grain multiprocessor architecture, based on an array of digital signal processors (DSPs), is presented to demonstrate the possibility of a parallel implementation of image-understanding algorithms. Aerial image understanding is investigated as an application. The system is designed to exploit temporal and spatial parallelism. A good speed-up was obtained for low- and intermediate-level operations. However, the speed-up for high-level operations was poor because of processor idle times, as the number of objects to be processed at higher level tends to be small. DSPs performed well as processing elements for number-crunching operations, but the performance was not so good for implementing symbolic operations.

Finite Difference Time-Domain (FDTD) is one of the most widely used numerical method for solving electromagnetic problems. Solutions for these problems are computationally expensive in terms of processing time. Recently, we develop a method, High Speed Low Order FDTD (HSLO-FDTD) that is proven to solve one dimensional electromagnetic problem with a reduction of 67% of processing time from the FDTD method. In this paper, we extend the method to solve two dimensional wave propagation problem. Since the problem is large, we develop the parallel version of HSLO-FDTD method on distributed memory multiprocessor machine using message-passing interface. We examine the parallelism efficiency of the algorithm by analyzing the execution time and speed-up.

Visibility computations, exhibiting a quadratic growth rate, are the bottleneck of high quality, real-time rendering of 3D models. In order to speed up visibility computations parallel or distributed-computing techniques can be used. Recent research demonstrates that scanline algorithms are suitable for balancing the workload and reducing the communication overhead of multiprocessor architectures. This paper summarizes known scanline algorithms and proposes a new one that takes time proportional to log and storage space proportional to in the worst case, where is the number of input line segments. The advantages of the proposed algorithm include that it can take real values obtained by the intersection of a 3D scene with the plane of the scanline as input and that it is optimal under several models of computation.

A scheduling algorithm is proposed for large-scale, heterogeneous distributed systems working on SPMD tasks with homogeneous input. The new algorithm is based on stochastic optimization using a modified least squares method for the identification of communication and performance parameters. The model of computation involves a server distributing tasks to clients. The goal of the optimization is to reduce execution time by the clients. The costs of getting the task from the server, execution of the task and sending the results back are estimated; and the scheduling is based on adaptive division of work (input for the clients) into blocks.

The visibility of a planar set of disjoint line segments, arising from the scanline approach to rendering three-dimensional scenes, is one of the classic problems in computer graphics. In order to solve the problem quickly, many authors proposed binary space partitioning (BSP) as a preprocessing, possibly breaking up the input line segments so that visibility is determined in time linear in the number of resulting segments. Tóth [Discrete & Comput. Geometry 30,1 pp. 3–16, 2003] demonstrated that a BSP may result in line segments. We demonstrate that the time and space complexities of the problem are and respectively, under the algebraic RAM model of computation. Introducing a more realistic model, a RAM with arbitrary-precision rational arithmetics, a deterministic algorithm is given that solves the problem directly, without the need of preprocessing, in time and space, regardless of the precision of the input data.

Ring signature is an important cryptographical tool to realize full anonymity. In many ring signature schemes, a member of the ring produced a signature , but his own cannot authenticate this signature was indeed produced by himself. To address it, in this work, we extend ring signature notion to present a novel scheme: ring signature scheme with self-authenticating which can authenticate that a member possesses a signature’s ownership. To improve efficiency of the scheme, our proposed scheme does not use function in Key Extract phase. In our proposed scheme, one pairing is needed in the signing phased and two pairings are needed in the verifying phase. Finally, we discuss the batch verification of our ring signature, and show that no matter how many signatures, only two pairing operators are needed in the batch verification. Thus, it is very efficient.

Recently many autogenous terrors are occurred in many countries, and this kind of terrors are spread mostly through the Internet. Therefore, there should be regulations and tools to monitor suspicious Internet activities. Recently, due to the increased number of cyber terrors, many countries try to enact regulations against web server misuses. But, the establishment of such laws is slowed by the concerns of privacy of web users.

This paper proposed legislative measures against cyber terrors as well as the proposed scheme to collect terror-related information from web servers. The result of this research can reduce overall costs of anti-terror activities in the Internet.

Security of an information system depends to a large extent on its ability to authenticate legitimate users as well as to withstand attacks of various kinds. Confidence in its ability to provide adequate authentication is, however, waning. This is largely due to the wrongful use of passwords by many users. In this paper, the design and development of keystroke pressure-based typing biometrics for individual user’s verification which based on the analysis of habitual typing of individuals is discussed. The combination of maximum pressure exerted on the keyboard and time latency between keystrokes is used as features to create typing patterns for individual users so as to recognize authentic users and to reject impostors. Support vector machines (SVMs), which is relatively new machine learning, is used as a pattern matching method. The effectiveness of the proposed system is evaluated based upon False Reject Rate () and False Accept Rate (). A series of experiment shows that the proposed system is effective for biometric-based security system.

DDoS attack methods become more sophisticated and effective. An attacker combines various attack methods, and as a result, attacks become more difficult to be detected. In order to cope with these problems, there have been many researches on the defense mechanisms including various DDoS detection mechanisms.

SVM is suitable for attack detection since it is a binary classification method. However, it is not appropriate to classify attack categories such as SYN Flooding attack, Smurf attack, UDP Flooding, and so on. Because of this weakness, administrator does not react against the attack timely. To solve this problem, we propose a machine learning model based on Multiple Support Vector Machines (MSVMs), and a new DDoS detection model based on Multiple Support Vector Machines (MSVMs). The proposed model enhanced attack detection accuracy and it classifies attack categories well when the proposed model detects the attacks.

It is well known the fact that turbo codes has better performance as the number of iteration and the interleaver size increases in the AWGN channel environment. However, as the number of iteration and the interleaver size are increased, it is required much delay and computation for iterative decoding. Therefore, it is important to devise an efficient criterion to stop the iteration process and prevent unnecessary computations and decoding delay. In this paper, we propose a new stop criterion for decoding turbo codes. It is verifying that the proposal iterative decoding stop criterion can be reduced the average iterative decoding number compared to conventional schemes with a negligible degradation of the error performance.