Seminar on Applied Mathematics
Utorak, 01.12.2009. u 14:15, Sala 301f, MI SANU:
Lav Ivanovic, LSI Corporation, Kalifornija, SAD
O ISKUSTVIMA U PRIMENI MATEMATIKE U INDUSTRIJI CIPOVA
Sadrzaj: U saopstenju ce biti dat detaljniji osvrt na DFM (Design for Manufacturing) probleme koji se srecu u proizvodnji cipova. Takodje cje biti reci i o problemima zastite podataka ukljucujuci DPI (Deep Packet Inspection). Zastita podataka se implementira hardverski i koristi najsavremenije kriptografske metode.
Utorak, 08.12.2009. u 14:15, sala 301f, MI:
Prof.dr Zoran Bojkovic, University of Belgrade, Senior Member IEEE
QUALITY OF SERVICE IN NETWORK MULTIMEDIA SYSTEMS
Abstract: Networked multimedia communication originated in a confluence of two technological trends. The first involved the development of multimedia computing, while the second involved advances in networking. Advances continue in both of these areas.The challenge of multimedia communications is to provide applications that integrate text,sound,image and video information and to do it in a way that preserves the ease of the use and interactivity. First of all ITU and MPEG applications from the integration, storage, streaming, retrieval and media distribution must be taken into account. Also, digital broadcasting, together with infrastructures are of the most interest. Mobile services and applications as well as research activities,such as usability, user interfaces, mobile access to data basis and agent technologies represent a modern area. In the network layer, the emphasis is on connection admission control, resource and bandwidth allocation, congestion control for multicast communications as well as traffic modelling.
The Internet new services generate a traffic explosion. Network traffic is more dynamic and complex as new services are added. Thus, adaptive techniques that can predict traffic variability and dynamically allocate resources will be very useful. Networks traffic is complex as it exhibits strong dependencies and self -similarity. Therefore classical models of time series such as Poisson and Markov processes are not appropriate for its modelling. Starting from this point, network traffic analysis (network engineering, traffic modelling, multimedia traffic management, resource and bandwidth allocation, congestion control) will be presented as an overview. This is of interest for a number of communication and networking tasks, such as designing, controlling, planning and maintaining a network, determining pricing schemes and QoS. Meeting QoS guarantees is an end-to-end issue, that is from application to application. It is essential for QoS to be configurable, predictable and maintainable systemwide, including the end-system devices, subsystems and networks. Thus, the second part of this work contains selection and configuration of QoS mechanisms, architecture, QoS from providers/'customers' viewpoints as well as IP oriented QoS.
Utorak, 15.12.2009. u 14:15, sala 301f, MI SANU:
Doc. dr Sasa Lazarevic, Faculty of Organizational Sciences, Belgrade
THE DEVELOPMENT OF NEURAL NETWORK IN FORM OF XML WEB SERVICE FOR SOLVING THE PROBLEMS OF DATA MINING'S CLASSIFICATION
Abstract: Various classification algorithms/tools became available due to a surge of interdisciplinary research interests in the areas of data mining and knowledge discovery. The approach proposed in this paper redefines the commonly used form of Neural Network as stand-alone software and proposed Neural Network as distributed software component. The objective is to expose a neuroclassifier in form of XML Web Service.
Key words and phrases: Data Mining and Data Base Modeling, Software for Data Analysis, Data Mining,Classification, Neural Network.
Utorak, 22.12.2009. u 14:15, sala 301f, MI SANU:
Dr. Ing. Tamara Nestorovic, Mechanik adaptiver Systeme, Ruhr-Universitat Bochum, Germany
MODEL DEVELOPMENT OF A SMART ACOUSTIC STRUCTURE AND ITS IMPLEMENTATION FOR THE MODEL-BASED ACTIVE STRUCTURAL ACOUSTIC CONTROL
Abstract: Two approaches for model development of a smart acoustic box are presented: the finite element (FE) approach and the subspace identification. Both approaches result in a state-space model, which can be used for obtaining the frequency responses and for the controller design. In order to validate the developed FE model and to perform the subspace identification, an experimental Hardware-in-the-Loop set-up with the acoustic box and dSPACE system is used. Based on the developed models the control methods are suggested and implemented for the noise reduction within the acoustic structure. As the noise measure, the pressure of the acoustic fluid at a prescribed point within the acoustic box is defined. Investigations through numerical simulations and experimental studies have shown a good agreement between the FE-based and identified models as well as a considerable reduction of the noise level using the suggested control methods (optimal and adaptive control).