Comparison of Service Discovery Protocols
M. Bratoev, B. Bontchev
Abstract
The advent of wireless, mobile and ubiquitous computing has made it necessary to develop a highly-dynamic infrastructure that enables devices to advertise their services and clients to easily locate and utilize a particular service out of hundreds of accessible services. There must be no need for complex configuration, administration or device driver installation. Service discovery is an important element in these intelligent computer networks and contributes for automated discovery, seamless information exchange and remote control between devices. This paper is represents a survey of several prominent service discovery technologies, compares their major features and outlines the challenges and trends in development of service discovery protocols during next years.
Cost Evaluation of Methods for Query Processing in Deductive Database Systems
V. Iltchev
Abstract
The paper describes: the important characteristics of an extensional database; a sample intensional database; some sample queries, whose processing will be evaluated; a cost metrics, based on the number of intermediate facts generated. On this basis, a cost comparison between well-known methods for query processing and a new bottom-up method, developed by the author, has been made.
Assets Selection Criteria for Portfolio Diversification
Based on Reward-to-Variability Ratio
G. Gatev, A. Malakhova
Abstract
Securities evaluation criteria are presented to select the assets for portfolio diversification. Modifications of the Sharpe coefficient are proposed, based on the new introduced risk measures. The effect of diversification is shown on the base of the introduced coefficients. The proposed selection criteria are evaluated according to the efficiency of a composed portfolio, measured by the Sharpe coefficient. Comparative analysis of the introduced securities evaluation criteria is held. Case study based on the Stock Exchange “Russian Trading System” is performed.
Efficiency of Parallel Combinatorial Search on Multicomputer Platforms
P. Borovska
Abstract
In this paper the efficiency of parallel combinatorial search on multicomputer platform has been investigated on the basis of MPI+OpenMP parallel program implementations. The parallel algorithms under consideration are parallel versions of backtrack search and branch-and-bound search for the case studies of the N Queens problem and Sam Loyd’s puzzle as benchmarks, respectively. Both parallel computational models are based on the manager/workers algorithmic paradigm implying dynamic strategy for load balancing in the parallel system. The performance parameters of the parallel system have been estimated and analyzed by means of parallelism profiling and benchmarking the hybrid parallel programs. The scalability of the parallel system has been investigated with respect to the parallel machine size and the parallel computational workload.
Algorithm for Face Detection, Adapted for Platforms with Limited Resources
R. Ivanov
Abstract
Face detection at unknown illumination is a complex process which does not have a simple solution. The task is additionally complicated when the used hardware platform is with limited processor power and memory size. This paper considers the algorithm realization of face detector that can be used for creation of Java applications for mobile terminals with built-in photo camera. The aim is creating of mass accessible application for people with impaired vision so that they can be informed (by Text To Speech conversion) about availability and number of faces within the frame, obtained by the photo-camera.
A System for Dynamic Simulation of Dressed Walking Characters
V. Dochev, T. Vassilev
Abstract
This paper presents a system for dressing virtual humans with dynamic simulation capabilities. The garments are constructed of cutting patterns, imported from apparel CAD systems. Two approaches for editing clothes were implemented and investigated: a standalone programme and a plug-in to Adobe Illustrator. The system produces dressed virtual humans using as an input a body file and a garment file. It is then possible to carry out simulations of walking dressed humans. For the latter a hybrid approach for cloth-body collision detection was developed. It is based on both image-and object-space techniques with the intention to incorporate their advantages. Images of both static and dynamic simulations produced with the system, along with performance data are given at the end of this paper.