Algorithmic mechanism design for load balancing in distributed systems

Computational grids are promising next-generation computing platforms for large-scale problems in science and engineering. Grids are large-scale computing systems composed of geographically distributed resources (computers, storage etc.) owned by self interested agents or organizations. These agents may manipulate the resource allocation algorithm in their own benefit, and their selfish behavior may lead to severe performance degradation and poor efficiency. In this paper, we investigate the problem of designing protocols for resource allocation involving selfish agents. Solving this kind of problems is the object of mechanism design theory. Using this theory, we design a truthful mechanism for solving the static load balancing problem in heterogeneous distributed systems. We prove that using the optimal allocation algorithm the output function admits a truthful payment scheme satisfying voluntary participation. We derive a protocol that implements our mechanism and present experiments to show its effectiveness.     

High voltage (>1kV) and high current gain (32) 4H-SiC power BJTs using Al-free ohmic contact to the base

This letter reports the design and fabrication of 4H-SiC bipolar junction transistors with both high voltage (>1kV) and high dc current gain (/spl beta/=32) at a collector current level of I/sub c/=3.83A (J/sub c/=319 A/cm/sup 2/). An Al-free base ohmic contact has been used which, when compared with BJTs fabricated with Al-based base contact, shows clearly improved blocking voltage. A specific on-resistance of 17 m/spl Omega//spl middot/cm/sup 2/ has been achieved for collector current densities up to 289 A/cm/sup 2/.     

Design of radiation-emission measurements of an air-traffic surveillance radar

The current work presents the procedure of designing and carrying out measurements of radiation emission from an air-traffic surveillance radar. The system requirements are first given, and then the equipment used is described. The steps to determine the operating characteristics of the radar are specified in detail. The necessary methods for manual power measurements and calculations are next explained. Since the final measurement system was fully automated, its main features (topology, protocol of operation) are also given. Finally, some measurement results taken during the initial phase of system operation are presented and discussed.     

Kinetic genesis of some statistic regularities of fracture

The statistics of strength and longevity of solids has been the subject of investigation for many years. Some common regularities in fracture statistics have been experimentally revealed. These are Jhurkov's exponential and power-type longevity equations, Weibull's statistics of strength, and different types of size effect. However, the present notions on the nature of the observed statistic regularities are far from being complete. Here the object is to examine fracture statistics on the basis of a previously suggested kinetic approach. We consider the evolution of the statistical ensemble of cracks under static and steady-rate loading of a solid. The growth of an individual crack is therewith treated as a spasmodic stochastic process and presented by a time-dependent distribution density of a crack size. That function is used to construct the conditional longevity distribution with respect to the growth of an individual crack with the prescribed initial size. The final longevity distribution is found as the distribution of a minimum of conditional longevities. That procedure has been realized for both brittle and quasi-brittle cases. The obtained longevity and strength distributions were found to be in qualitative agreement with the observed statistic regularities. However, the conventional `weak link concept' and proposals for the statistics of dangerous defects are in conflict with the obtained results. It is moreover shown that, within the limits of the wide-spread concept, Jhurkov's longevity equation and Weibull's statistics of strength are mutually exclusive. Although the results were obtained from simple models, there is a good probability of kinetic origin for the basic statistic regularities of fracture.     

Weyl filters for linear array data

The Weyl transform, which is closely connected to the Wigner distribution, allows one to reconstruct (temporally and spatially) localized frequency-wavenumber spectral components. Such filters were suggested in time-frequency analysis by Kozek and Hlawatsch (1992). The theory can be naturally extended to signals recorded in space and time. Thereby, wavefield components with predefined varying spectral properties like, for example, wavefront curvature, can be synthesized. The numerical implementation of such filters is in the most general case a computationally expensive task. We present a simplified implementation, which assumes a specially shaped filter pass region. The assumption resembles a ray approximation and allows one to track a temporally and spatially unidimensional wavefield recording for curved wavefronts in reasonable computation times. The suitability of the method is demonstrated by synthetic data and seismic field data, although it is quite general and not restricted to seismic applications.     

An optimum parallel architecture for high-speed real-time digital signal processing

The authors describe a parallel processing architecture for real-time digital signal processing that has demonstrated virtually 100% data processing efficiency in a number of areas. The Teamed-Architecture Signal Processor (T-ASP) is a field-proven, commercially available optimal system solution to the extremely high computational and I/O rates encountered in modern digital-signal-processing environments. The design of T-ASP involves the consideration and implementation of many architectural concepts used to enhance the performance of a computer, including programmability, parallel processing, vector processing and pipelining, memory interleaving, double cache memories, multiple high-speed I/O interfaces, and segmentation of the processors for elimination of both CPU and data-handling overhead. The authors discuss hardware architecture design and implementation; hardware management; and software architecture design and implementation.<>