|
IcETRAN-a 2014 PROGRAM
ACOUSTICS
AKI1 Audio signals, Electroacoustics, Psychological
acoustics
Chair: Dragana Šumarac Pavlović, School of
Electrical Engineering, Belgrade, Serbia
Thursday, June 5, Room 2, 8:30
AKI1.1
THE LPCC-DTW ANALYSIS FOR WHISPERED SPEECH RECOGNITION
Branko R. Marković, Technical College, Computing and
Information Technology Department, Čačak, Serbia
Đorđe T. Grozdić, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
In this paper we explained the results of recognition for
whisper and normal speech using the LPCC analysis and DTW method. The
isolated words used for this experiment are from the Whi-Spe database. Two
subsets of this database which contains words of colours and words of numbers
are taken in consideration. The speakers are 5 males and 5 females. The
vectors for comparison are based on LPCC, delta and delta-delta coefficients.
For a local constraint Type I is applied at DTW method without constraints on
the global path. The results are presented in the form of tables and a
diagram.
AKI1.2
AUDIO SIGNAL DE-NOISING USING WAVELETS ALGORITHM IN MATLAB
AND LABVIEW
Đorđe
Damnjanović, Faculty of Technical Sciences Čačak,
University of Kragujevac, Čačak,
Serbia
Milan
Gojković, Faculty of Electronic Engineering, University of Niš,
Niš, Serbia
Dејan Ćirić, Faculty of
Electronic Engineering, University of Niš, Niš, Serbia
This paper describes usage of wavelet algorithms in process
of de-noising of audio signals. Usage of wavelets nowadays is widespread and
wavelet algorithms are very interesting for solving some engineering
problems, such as noise cancellation in audio and image signals. Applications
which are using the Fourier transform algorithms can be formulated using
wavelets to provide more accurately localized temporal and frequency information.
Matlab and LabVIEW program packages are very useful in these areas, first of
all for analysis these signals, and later for some further processing such as
de-noising. Both packages have built-in functions that can be easily used for
solving these problems.
AKI1.3
ANALYSIS OF DIRECT SOUND AND FIRST REFLECTIONS USING
SPHERICAL MICROPHONE ARRAY
Ana
Đorđević, Faculty of Electronic Engineering, University
of Niš, Niš, Serbia
Dejan
Ćirić, Faculty of Electronic Engineering, University of Niš,
Niš, Serbia
Marko
Ličanin, Faculty of Electronic Engineering, University of Niš,
Niš, Serbia
Room impulse response represents a main characteristic for
analysis of rooms as acoustic systems. However, it provides very limited
information about spatial properties of the sound field, when measurements
are performed with a single measurement microphone. Recently developed
spherical microphone arrays have been used for analyzing the spatial
properties of sound field. When such a device is applied for a room impulse
response measurement, it is able to provide additional (spatial) information.
This paper presents an investigation of direct sound and first reflections of
the impulse responses using 20 sample points spherical microphone array.
Here, floor reflection is considered, but also 90 and 180 degrees reflections
from a wooden panel in an anechoic chamber. These results are compared with
the impulse responses measured by the recently built 32 sample points
spherical microphone array in a non-anechoic environment.
ANTENNAS AND PROPAGATION
API1. Antennas and
Propagation
Chair: Branko
Kolunžija, School of Electrical Engineering, Belgrade, Serbia
Monday, June 2, Room 4, 14:00
API1.1
CONVERGENCE OF THE HIGHER ORDER TIME-DOMAIN FINITE ELEMENT METHOD
IN MODELING OF 1-D ELECTROMAGNETIC PROBLEMS
S. V. Savić, University of Belgrade, School of
Electrical Engineering, Belgrade, Serbia
M. M. Ilić, University of Belgrade, School of Electrical
Engineering, Belgrade, Serbia and Colorado State University, Department of
Electrical and Computer Engineering, Fort Collins, USA
We investigate the convergence of a higher order and
large-domain finite element method (FEM) in direct electromagnetic modeling
of 1-D electromagnetic problem in the time domain (TD). In the process, we
vary the parameters of the time-domain discretizations within the numerical
models and the number of time-stepping samples in the time marching schemes.
The numerical results are verified by comparison with analytical solutions
and by accurate frequency-domain FEM solutions, providing valuable insight
into accuracy and convergence properties of the higher order TDFEM for
different time marching schemes.
API1.2
EDDY CURRENT POWER LOSS IN THE SHIELD OF AN N-CONDUCTOR TRANSMISSION LINE
Dragan Filipović, Faculty of Electrical Engineering,
University of Montenegro, Bul. Dzordza Vasingtona bb, 81000
Podgorica, Montenegro
Tatijana Dlabač, Maritime
Faculty, University of Montenegro, Dobrota 36, 85330 Kotor, Montenegro
In this paper a simple formula for the eddy current power
loss in the shield of an N‑conductor
transimission line is derived under the assumption that the shield and the
conductors are thin. The general theory is illustrated by a particular case
of a shielded symmetrical three‑phase line.
API1.3
CIRCULARLY
POLARIZED PRINTED ANTENNA WITH TUNABLE ELEVATION ANGLE
Ivana Radnović, IMTEL Communications a.d, Bulevar
Mihajla Pupina 165b, 11070 Belgrade, Serbia
Aleksandar Nešić, IMTEL Komunikacije a.d, Bulevar
Mihajla Pupina 165b, 11070 Belgrade, Serbia
Dušan Nešić, ICTM-CMT, University of Belgrade, Njegoševa
12, 11000 Belgrade, Serbia
The paper presents the antenna structure with circular
polarization and possibility of tuning the elevation angle i.e. obtaining maximum
radiation at angles declined from the broadside direction, which is
particularly interesting in communications between mobile objects and
satellites. The antenna consists of two printed crossed dipoles, feed line
and a mechanism for adjusting the height of the antenna in regard to vehicle
roof. Simulation and optimizations have been carried out using program
package WIPL-D and analysis of obtained results proves the validity of
presented concept. Simulated gain is between 4.3 dBi and 7.3 dBi, depending
on the antenna position; axial ratio (AR) is less than 3 dB in the
operational range of frequencies (2%); reflection coefficient S11 is less
than -19 dB. Antenna is designed for INMARSAT frequency range.
API1.4
COMPARATIVE PERFORMANCE ANALYSIS OF NLMS AND VSS LMS
ALGORITHM FOR PLANAR ANTENNA
Luka Lazović, Student, Faculty of Electrical Engineering, University of Montenegro,
Podgorica, Montenegro
Ana Jovanović, Faculty of Electrical Engineering, University of Montenegro,
Podgorica, Montenegro
Vesna Rubežić, Faculty of Electrical Engineering, University of Montenegro,
Podgorica, Montenegro
This paper presents a comparative performance study of NLMS
(Normalized Least Mean Square Algorithm) and VSS LMS (Variable Step Size LMS)
adaptive algorithm. A smart antenna system with innovative signal processing
can enhance the resolution of a beamformer. Super resolution algorithms take
advantage of array antenna structures to better process the incoming signals
and they also have the ability to identify multiple interferer signals and
attenuate them. Adaptive beamforming is achieved using the LMS algorithm and
its variation for directing the main beam towards the desired source signals
and generating deep nulls in the directions of interfering signals. We
studied the effect of antenna array parameters in terms of its size and
element spacing. In simulations we study the influence of angle of arrival on
algorithm behavior and MSE (Mean Square Error).
AUTOMATIC CONTROL
AUI1. Systems parameter
and state estimation
Chair: Milorad Božić, University of Banja
Luka
Monday, June 2, Room 2, 14:00
AUI1.1
ROBUST ADAPTIVE PARAMETER ESTIMATION OF THE THERMAL POWER
PLANT COMBUSTION PROCESS
Aleksandra Marjanović, School of Electrical Engineering,
University of Belgrade, Serbia
Sanja Vujnović, School of Electrical Engineering,
University of Belgrade, Serbia
Veljko Papić, School of Electrical Engineering,
University of Belgrade, Serbia
Predrag Todorov, School of Electrical Engineering, University
of Belgrade, Serbia
Optimization of the combustion process control, as one
highly complex system, represents a great challenge for the engineering
scientific community. Proper regulation of the overall combustion process can
significantly increase safety and energy efficiency, while decreasing the
pollution in the system. In order to propose an adequate control algorithm,
it is necessary to conduct a proper analysis and to obtain a reliable model
of the system. The paper describes modeling of a firing system using the
measurements obtained from the temperature visualization system, which is
installed in the 350MW boiler of the Nikola Tesla Power Plant, in Obrenovac,
Serbia. We propose an application of a robust adaptive parameter estimation
method, based on a variable forgetting factor onto such a system and provide
evaluation of this identification procedure.
AUI1.2
A NEW ADAPTIVE ROBUSTIFIED PREDICTION ALGORITHM WITH UNKNOWN
NOISE STATISTICS
Dunja Đurović, Faculty of Informatics and
Computing, Singidunum University, Serbia
Jelena Gavrilović, Faculty of Technical Sciences,
Singidunum University, Serbia
Ivana Kostić Kovačević, Faculty of Informatics
and Computing, Singidunum University, Serbia
Branko Kovačević, School of Electrical Engineering,
University of Belgrade, Serbia
A new robust adaptive predictor for situations where noise
statistics are not fully known is presented in the paper. First an optimized
predictor is developed, based on the minimization of a generalized mean
square prediction error. It determines the structure of the robust adaptive
predictor, which is synthesized through minimization of a modified criterion
in which a quadratic function is introduced instead of an arbitrary
non-linear function of the prediction error. The non-linear function is
determined by applying Huber’s min-max approach, which assumes a priori
knowledge of the distribution class to which the actual unknown noise
distribution belongs. The resulting non-linearity is a maximum likelihood
criterion function, and is determined by the least favorable probability
density function within the given class, which carries minimal information
about the estimated parameters. Unknown parameters of the predictor are
estimated at each step by applying a recursive algorithm of the stochastic
gradient type.
AUI1.3
FEATURE EXTRACTION FOR EMOTION CLASSIFICATION FROM SPEECH
SIGNAL
Milana Milošević, School of Electrical Engineering,
University of Belgrade, Serbia
Željko Đurović, School of Electrical Engineering,
University of Belgrade, Serbia
Emotional speech recognition is one aspect of behavioral
human analysis as a part of Social Signal Processing as a new
multidisciplinary area for which there is an increasing interest of
scientist. This paper is a review of basic elements of emotion analysis in
speech signal which are used in previous researches: which characteristics
are most commonly used, which databases are used and in a which way test
sequences are prepared and recorded, and which classificators are used and in
a which way. Successful emotion recognition is on the level of human
recognition which is about 75%, but under special conditions can reach over
90%.
AUI1.4
MULTIVARIATE MEDIANS AND HALFSPACE DEPTH: ALGORITHMS AND
IMPLEMENTATION
Milica Bogićević, School of Electrical Engineering,
University of Belgrade, Serbia
Milan Merkle, School of Electrical Engineering, University of
Belgrade, Serbia
We are considering the notions, properties and algorithms’
implementations of data depth which represents the median of higher dimensional
data. Our main objective is to present the snapshot of the data depth with
respect to half-space depth, also known as location depth or Tukey depth.
Although the problem is NP-hard, there are ways to compute nontrivial lower
and upper bounds of the depth. Computation of Tukey depth is very demanding
and even for low dimension dataset, it requires all one dimensional
projections to be considered. This is the reason why implementations of
particular algorithms represent a challenge, not only in order to calculate
deepest data location, but also in order to visualize initial data set and
its calculated results.
AUI1.5
SENSOR FAULT DIAGNOSIS IN NONLINEAR PLANTS: A MARGINALISED
PARTICLE FILTER APPROACH
Predrag Tadić, School of Electrical Engineering,
University of Belgrade, Serbia
Goran Kvaščev, School of Electrical Engineering,
University of Belgrade, Serbia
Željko Đurović, School of Electrical Engineering,
University of Belgrade, Serbia
Branko Kovačević, School of Electrical Engineering,
University of Belgrade, Serbia
We consider the problem of detecting, isolating and
reconstructing sensor faults in plants described by general, nonlinear,
non-Gaussian state space models. The faults are modelled as additive
time-varying terms in the measurement equation. The model is augmented by
including the faults in the state vector. Particle filtering is then employed
to simultaneously estimate both the original states and the faults in real
time. We take advantage of the partial conditional linearity of the augmented
model through Rao-Blackwellisation, thus significantly reducing the otherwise
high computational cost. Computer simulations on the benchmark three tank
process testify to the effectiveness of the proposed algorithm.
AUI1.6
A NEW APPROACH TO ADAPTIVE NON-STATIONARY SIGNALS PARAMETER
IDENTIFICATION
Slobodan Drašković, School of Electrical and Computer
Engineering of Applied Studies, Belgrade, Serbia
Goran Kvaščev, School of Electrical Engineering,
University of Belgrade, Serbia
Vera Petrović, School of Electrical and Computer
Engineering of Applied Studies, Belgrade, Serbia
Željko Đurović, School of Electrical Engineering,
University of Belgrade, Serbia
Branko Kovačević, School of Electrical Engineering,
University of Belgrade, Serbia
Modeling of non-stationary signals can be achieved through
autoregressive models using adaptive recursive least squares technique with
variable forgetting factor. In this paper, a new method for adaptation of the
forgetting factor based on absolute finite differences is presented. The
method has good tracking ability of the non-stationary parts of the signal,
and satisfactory low bias and variance in stationary ones. The validation of
the approach is presented with simulations.
AUI1.7
MOVING HORIZON STATE ESTIMATION: MULTI-STEP CONSENSUS SCHEME
Vukašin B. Ćirović, School of Electrical
Engineering, University of Belgrade, Serbia
Srđan S. Stanković, School of Electrical
Engineering, University of Belgrade, Serbia
Moving horizon state estimation (MHE) is becoming an important
tool for state estimation. Recent papers focus on distributed state
estimation over sensor networks. In this case the system state could be
estimated by two or more sensors, thus, a consensus algorithm may be used to
define a common state estimate. In this paper, a review on the basics of MHE
algorithms is presented. Attention is focused on distributed moving horizon
estimation and current problems. An improvement of the existing MHE algorithm
for distributed systems based on a multi-step consensus scheme is proposed
and the obtained results are discussed. Finally, open problems are listed
with the aim to prepare a basis for future research.
AUI1.8
BLIND MACRO CALIBRATION FOR SENSOR NETWORKS IN STOCHASTIC
ENVIRONMENT
Miloš Stanković, Innovation Center, School of Electrical
Engineering, University of Belgrade, Serbia
Srđan Stanković, School of Electrical Engineering,
University of Belgrade, Serbia
Karl Henrik Johansson, ACCESS Linnaeus Center, School of
Electrical Engineering, KTH Royal Institute of Technology, Stockholm, Sweden
In this paper a novel distributed algorithm forblind
macro-calibration in sensor networks based on outputsynchronization is
proposed. The algorithm is formulated asa set of gradient-type recursions for
estimating parametersof sensor calibration functions. It is proved, on the
basis ofan originally developed methodology for treating
higher-orderconsensus schemes, that the algorithm achieves
asymptoticagreement for sensor gains and offsets in the mean square senseand
with probability one in the case of additive measurementnoise, either using
the a priori knowledge of noise covariance ora modification of the original
algorithm based on instrumentalvariables. Special attention is paid to the
situation when a subsetof sensors remain with fixed characteristics. An
illustrativesimulation example is provided.
AUI2. Control of complex systems
Chair: Branko Kovačević, University of
Belgrade, Serbia
Monday, June 2, Room 2, 16:00
AUI2.1
FLEXIBLE CASCADE CONTROL FOR TRAJECTORY TRACKING OF A
QUADROTOR
Ivan Petruševski, School of Electrical Engineering,
University of Belgrade, Serbia
Aleksandar Rakić, School of Electrical Engineering,
University of Belgrade, Serbia
This paper presents a cascade control approach for
trajectory tracking of a four rotor helicopter called Quadrotor. Inner
attitude control is proposed in the form of separate PD controllers, while
the backstepping approach is utilized to obtain appropriate outer trajectory
tracking control. Dynamic quadrotor model is presented and used in the design
of control algorithm. Various simulations were performed implying the
proposed control strategy can be implemented as control algorithm of fully
autonomous quadrotor.
AUI2.2
AN APPROXIMATE INTERNAL MODEL-BASED NEURAL CONTROL FOR
NONMINIMUM PHASE SYSTEMS
Jasmin Igić, mtel a.d., Banja Luka, Bosnia and
Herzegovina
Milorad Božić, Faculty
of Electrical Engineering, University of Banja Luka, Bosnia and
Herzegovina
An Approximate Internal Model-based Neural Control (AIMNC),
using Multi Layer Perceptron (MLP) neural networks is proposed for nonminimum
phase systems. In the proposed control strategy only one neural network,
which is the neural model of the plant, should be trained off-line. An
inverse neural controller can be directly obtained from the neural model.
Simulations demonstrate the satisfactory performance of the proposed AIMNC
strategy and confirms that the system provides zero steady-state error in
case of the constant reference and constant disturbances.
AUI2.3
ITERATIVE LEARNING FEEBACK CONTROL FOR SINGULAR FRACTIONAL
ORDER SYSTEM- PDA TYPE
Mihailo Lazarević, Faculty of Mechanical Engineering,
University of Belgrade, Serbia
Panagiotis Tzekis, Electronic Engineering Department, School
of Technological Applications, A.T.E.I Thessaloniki, Greece
In this paper an closed-loop PDalpha type iterative
learning control (ILC) of fractional order singular system.
Particularly, we discuss fractional order linear singular systems in state
space form. Sufficient conditions for the convergence of a proposed
PDalpha type of learning control algorithm for a class of fractional
order singular system are given in time domain. Finally, a simulation example
shows the feasibility and effectiveness of the approach.
AUI2.4
IMPLEMENTATION OF ACTIVE DISTURBANCE REJECTION CONTROL ON
FPGA
Momir Stanković, Military Academy, University of
Defence, Belgrade, Serbia
Stojadin Manojlović, Military Academy, University of
Defence, Belgrade, Serbia
Slobodan Simić, Military Academy, University of Defence,
Belgrade, Serbia
Zoran Jovanović, Faculty of Electronic Engineering,
University of Niš, Serbia
Implementation of Active Disturbance Rejection Control
(ADRC) on Field Programmable Gate Array (FPGA) was presented in this paper.
Discrete ADRC for speed servo system with predictive and current extended
state observer (ESO) was derived and compared by simulation. A hardware
design of ADRC using high-level blocks and system-level hardware design
tools, actually Xilinx’s System Generator™ is presented. The experimental
results largely agree with simulation and maintain good system’s performance
regardless different working conditions with external disturbance.
AUI2.5
OPTIMAL CONTROL OF SINGULAR SYSTEMS: CONTINUOUS TIME CASE
Nataša A. Kablar, Faculty of Computer Science and Lola
Institute, Belgrade, Serbia
Vladimir Kvrgić, Lola Institute, Belgrade, Serbia
Dragutin Debeljković, Faculty of Mechanical
Engineering, University of Belgrade, Serbia
In this paper for the class of singular dynamical systems
we present optimal control results. We develop unified framework for feedback
optimal and inverse optimal control involving a nonlinear-nonquadratic
performance functional. It is shown that the cost functional can be evaluated
in closedform as long as the cost functional considered is related in a
specific way to an underlying Lyapunov function that guarantees asymptotic
stability of the nonlinear closed-loop singular system. Furthermore, the
Lyapunov function is shown to be a solution of a steady-state,
Hamilton-Jacobi-Bellman equation.
AUI2.6
OPTIMAL CONTROL OF SINGULAR SYSTEMS: DISCRETE TIME CASE
Nataša A. Kablar, Faculty of Computer Science and Lola
Institute, Belgrade, Serbia
Vladimir Kvrgić, Lola Institute, Belgrade, Serbia
Dragutin Debeljković, Faculty of Mechanical
Engineering, University of Belgrade, Serbia
In this paper for the class of discrete time singular
dynamical systems we present optimal control results. We develop unified
framework for feedback optimal and inverse optimal control involving a
nonlinear-nonquadratic performance functional. It is shown that the cost
functional can be evaluated in closed-form as long as the cost functional
considered is related in a specific way to an underlying Lyapunov function
that guarantees asymptotic stability of the nonlinear closedloop discrete
time singular system. Furthermore, the Lyapunov function is shown to be a
solution of a steady-state, Hamilton-Jacobi-Bellman equation.
AUI2.7
PERMANENT MAGNET DC MOTOR ADDITIVE FAULTS DETECTION AND
ISOLATION
Sanja Antić, Faculty of Technical Sciences
Čačak, University of Kragujevac, Serbia
Željko Đurović, School of Electrical Engineering,
University of Belgrade, Serbia
This paper presents methods for detection and isolation of
four selected additive faults of DC (direct current) motor. The motor model
is nonlinear due to temperature-dependent coefficient of viscous friction.
Nevertheless, it is shown that linearized set of residuals provides adequate
detection and fault isolation. Residuals were designed using two fundamental
residual enhancement approaches: structural residuals and directional
residuals. It has been shown that the structural residuals can be efficiency
designed for all of the four faults using alternative interpretations with
series of eliminations. On the other hand, the number of faults for which
independent directional responses can be design is limited to the number of
system outputs that is two. So in order to detect and isolate all of the four
faults it is necessary to design directional residuals in two steps: firstly
with assigned independent response directions for the first two faults,
secondly with assigned independent response directions for the last two
faults. Experiments have been performed on a PM DC motor with Compact RIO
9074 real-time processor and its NI C Series I/O modules.
AUI2.8
A MISSILE GUIDANCE METHOD BASED ON A SLIDING MODE CONTROLLER
Ali AmerAl_Rawi, School of Electrical Engineering, University
of Belgrade
Stevica Graovac, School of Electrical Engineering, University
of Belgrade
Badlisha Ahmad,
AbidYahya, School of Electrical Engineering, University of Belgrade
In this paper, a modified guidance and control sliding mode
controller (MGCSMC) method is proposed in a guided missile system. The
modified sliding mode controller (MSMC) algorithm is adopted to enable the
missile to reach the desired target within a short period of time. The target
always makes high manoeuvres when the missile is close it. This issue has
been treatment in Guidance and control (G&C) by using a MSMC instead of
the traditional method such proportional navigation method (PN).Theoretical
analysis is conducted to reduce the miss-distance and chattering phenomenon
in SMC. Simulation of MGCSMC compared
with PN method show an improvement of about 80%, 47% and 20%for the
chattering, miss-distance and finite time, respectively. Furthermore, for the
high-altitude target, the MGCSMC improves the acceleration and flight angle
of the missile by approximately 65%., and achieve100% accuracy, whereas in PN
method achieves only 60% accuracy under the same conditions.
POWER ENGINEERING
Session EEI1. Power Engineering
Chair: Dejan
Reljić, Faculty
of Technical Sciences, University of Novi Sad, Serbia
Wednesday, June 4, Room 2, 8:30
EEI1.1
ON-LINE AIRGAP ECCENTRICITY FAULT DETECTION IN A THREE-PHASE
INDUCTION MOTOR
Dejan Reljić,
Faculty of Technical Sciences, University of Novi Sad, Serbia
Josif Tomić, Faculty
of Technical Sciences, University of Novi Sad, Serbia
Željko Kanović, Faculty of Technical Sciences, University of Novi Sad, Serbia
In this paper, a method for the on-line detection of the
airgap eccentricity fault in a three-phase cage induction motor has been
proposed. The method is based on a Motor Current Signature Analysis (MCSA)
approach, a technique that is often used for induction machine condition
monitoring and fault diagnosis. It is based on the spectral analysis of the
stator line current signal and the frequency identification of specific
harmonic components, which are created as a result of motor faults. However,
in this paper MSCA has been proposed as a tool for the on-line detection of
abnormal levels of airgap eccentricity. The most commonly used method for
current signal spectral analysis is based on the Fast Fourier transform
(FFT). However, due to the complexity and memory demands, the FFT algorithm
is not always suitable for real-time systems. Instead of the whole spectrum analysis,
this paper suggests only the spectral analysis on the expected airgap fault
frequencies employing the Goertzel’s algorithm. The proposed MCSA method with
the Goertzel’s algorithm allows continuous the real-time monitoring of airgap
eccentricity in a three-phase cage induction motor under the steady-state
condition. In this way airgap eccentricity can be detected at an early stage,
before serious failures occur, and costly secondary deterioration can be
avoided. The proposed technique has been conducted on both the line-connected
and the inverter-fed three-phase four-pole cage induction motor operated
under various load torques. An experimental low-cost data acquisition card,
which runs under LabView software, has been used to validate the algorithm.
EEI1.2
DYNAMIC MODEL OF SURFACE MOUNTED PERMANENT MAGNET SYNCHRONOUS
MACHINE
Gojko Joksimović, Faculty of Electrical Engineering in
Podgorica, University of Montenegro, Montenegro
Thomas M. Wolbank, Institut für Energiesysteme und
Elektrische Antriebe, Technische Universität Wien, Austria
Goran Stojčić, Institut für Energiesysteme und
Elektrische Antriebe, Technische Universität Wien, Austria
Ana Zogović, Faculty of Electrical Engineering in
Podgorica, University of Montenegro, Montenegro
In this paper, an approach to modeling of surface mounted
permanent magnet synchronous machine using winding function approach is
given. The specific machine with radial magnetized NdFeB permanent magnets is
modeled. Slotting effect from stator side and core saturation is neglected.
Complete set of equations that describes the machine is derived in detail.
Manner of self and mutual inductance between stator phase windings is
presented, too. Manner of calculation of magnetic flux linkages of stator
phase windings due to the rotating permanent magnets from rotor side as well
as electromagnetic torque calculation is also presented. The model was
verified in a manner that some transient regimes of the machine are modeled.
The changes in rotor speed, developed electromagnetic torque, stator currents
as well as back emf in stator phase windings are in line with expectations
for this type of machine.
EEI1.3
HARMONIC POWER FLOW STUDIES FOR RADIAL ELECTRIC DISTRIBUTION
SYSTEMS
Hani M. Baniodeh, Electrical & Electronics Department,
Faculty of Engineering, University of Sirte, Libia
Abdaslam Eajal, Electrical & Electronics Department,
Faculty of Engineering, University of Sirte, Libia
Saad Muftah Zeid, Electrical & Electronics Department,
Faculty of Engineering, University of Sirte, Libia
The Power flow studies are the backbone of the power system
analysis and design. These studies are normally carried out at the
fundamental frequency. However, due to the proliferation of electronic
devices in distribution systems, the power flow studies at harmonic
frequencies need to be conducted and that’s is for several reasons, First,
the system real power loss in a harmonic-rich environment consists of two
components, the fundamental component, the real power loss obtained at the
fundamental frequency, and the harmonic component, the real power loss
obtained at the harmonic frequencies. Second, the harmonic distortion level
at each bus needs to be determined in order to identify the harmonic sources
in the system. Thus, a harmonic power flow algorithm (HPF) for radial
distribution systems (RDSs) is developed in this project. The HPF algorithm
is based on a well known technique called a backward-forward sweep technique.
The backward sweep is employed to calculate the harmonic line section
currents, while the forward sweep is utilized to obtain magnitudes and phase
angles of the harmonic bus voltages. The HPF algorithm is fast compared with
the conventional power flow programs since there is no need to build up and
find the inverse of the bus admittance matrix or the Jacobian matrix which is
a time-consuming procedure. The HPF algorithm is tested on a 10-bus radial
distribution test system. The results obtained demonstrate the accuracy and
efficiency of the proposed HPF technique in carrying out on- and off-line
power system studies.
ELECTRIC CIRCUITS AND SYSTEMS AND SIGNAL PROCESSING
EKI1. Filter Analysis and Design and Advanced Data
Processing
Chair: Vlastimir Pavlović, Faculty of Electronic
Engineering, University of Niš, Serbia
Thursday, June 5, Room 4, 10:30
EKI1.1
ANALYSIS OF ATYPICAL FILTER STRUCTURES IN MATLAB
Aleksandar Radonjić, Crnogorski Telekom A.D., Podgorica,
Montenegro
Jelena D. Ćertić, School of Electrical Engineering, University of
Belgrade, Serbia
In this paper, we present a method of analysis of
specific/atypical filter structures in commercially available software
packages. As an example, we use an approximately linear phase half-band IIR
filter and MATLAB FDA Tool. We consider this as a good example because
current version of FDA Tool does not support directly design and analysis of
proposed structure. We present developed MATLAB script that creates filter
objects for the specific filter implementations and test those objects in FDA
Tool performing analysis of the quantization effects. We show that FDA Tool
can be used for the analysis of the filters even in the case when specific
design is not supported directly.
EKI1.2
CHARACTERISTICS OF NOVEL DESIGNED CLASS OF CIC FIR FILTER FUNCTIONS
OVER CLASSICAL CIC FILTERS
Vlastimir D. Pavlović, University of Niš, Faculty
of Electronic Engineering, Department of Electronics, Niš, Serbia
Dejan N. Milić, and Biljana P. Stošić, University of Niš, Faculty
of Electronic Engineering, Department of Telecommunications, Niš, Serbia
Novel class of selective multiplierless
Cascaded-Integrator-Comb (CIC) finite impulse response (FIR) filter functions
with improved frequency response characteristics is designed by use of CIC
sections of different lengths and presented here. The superiority of the
novel class of CIC filter functions is verified by comparing their
characteristics with those of the classical CIC filters under fair
conditions: the same number of cascaded sections and the same group delays.
EKI1.3
QUASIELLIPTIC TRANSFER FUNCTIONS WITH DOUBLE POLES
Maja M. Lutovac, Lola Institute, Belgrade, Serbia
Miroslav D. Lutovac, Singidunum University, Belgrade, Serbia
The largest pole Q factor of the second order transfer
functions determines the filter properties. With the larger Q factor we can
expect more deviations to component imperfections and more problems with the
dynamic range of active filters. There are a large number of methods for
reduction of Q factors, but the double critical poles are the most efficient
method. In this paper we provide systematic procedure for the design of
active filters with double poles that is based on elliptic rational
functions. The number of transfer function zeros is maximal with equiripple
stop-band attenuation, and with the half of the ripples in the pass-band with
respect to the stop-band.
EKI1.4
COMBINED TV FILTERING METHOD AND CS SIGNAL RECONSTRUCTION
Marijana Šćekić,
Radomir Mihajlović,
Andjela Draganić,
Irena Orović,
The procedure that combines Total Variation filtering
method and Compressive Sensing signal reconstruction is proposed in this
paper. Recently, Compressive Sensing has been intensively studied as a method
for signals acquisition. It has been shown that signals can be reconstructed
by using just a small set of random samples. However, the signal
reconstruction may not be efficient in the presence of noise. Therefore, we
considered a combined approach that performs Total Variation filtering prior
to Compressive Sensing reconstruction, in order to provide high accuracy of
reconstruction results. The procedure is tested on signals that appear in
wireless communications. The experiments demonstrate that the Total Variation
procedure successfully eliminates the Gaussian noise, while the filtered
signal can be successfully recovered using only 30% of signal samples..
EKI1.5
PROGRAMMING OF VICKERY AUCTION USING COMPUTER ALGEBRA SYSTEM
Miroslav D. Lutovac, Singidunum University, Belgrade, Serbia
Aleksandra M. Lutovac, Faculty of Economic, University of Belgrade, Belgrade,
Serbia
The market for information and communication networks and
mobile services is constantly growing. Technological progress creates new
challenges and new operators and service providers, with largely independency
of choosing the business models. It is
important immediately to recognize the most efficient business model, which
is more based on the one step e-auction process with unknown number of
auctioneers. One of the most popular auction models is Vickery auction, in
which the highest bidder wins but pays the second-highest bid. This variation
over the normal bidding procedure is supposed to encourage bidders to bid the
largest amount they are willing to pay, but the payoff can be very low if the
bidding prices is too high. In this paper, the efficient programming
procedure based on game theory and computer algebra systems is presented.
EKI1.6
AUTOMATIC LICENSE PLATE RECOGNITION USING CROSS CORELATION
Ivan Božić, School of Electrical Engineering, University
of Belgrade, Belgrade, Serbia
Ivan Lazić, ICT College, Belgrade, Serbia
Stefan Đinđić, School of Electrical Engineering, University of
Belgrade, Belgrade, Serbia
Maksim Đurđevac, Faculty of Mathematics, University of Belgrade,
Belgrade, Serbia
Marko Dragoslavić, School of Electrical Engineering, University of
Belgrade, Belgrade, Serbia
Automatic License Plate Recognition system is a real time
embedded system which automatically recognizes the license plate of vehicles.
There are many applications ranging from complex security systems to common
areas and from parking admission to urban traffic control. Automatic license
plate recognition has complex characteristics due to diverse effects such as
of light and speed. Proposed method is based on Hough transform, and
artificial neural networks.
EKI1.7
COMPARATIVE ANALYSIS OF MULTIFRACTAL PROPERTIES OF H.264 AND
MULTIVIEW VIDEO
Amela Zeković, School of Electrical and Computer
Engineering of Applied Studies, Belgrade, Serbia
Irini Reljin, School of Electrical Engineering, University of
Belgrade, Belgrade, Serbia
In this paper, a comparative analysis of multifractal
properties of a H.264 video and a multiview video is presented. Analyses of
the different views of the multiview video are performed, as well as analyses
of different streaming approaches of the multiview video, and results are
compared with those for the H.264 video. Calculation of the multifractal
properties is performed by the histogram method. For the analysis, publicly
available long frame size traces are used.
EKI2. Systems for data processing
Chair: Irini Reljin, School of Electrical Engineering,
University of Belgrade, Serbia
Thursday, June 5, Room 4, 12:00
EKI2.1
AN EXAMPLE OF COMPUTER MODELING OF MATCHED CRYSTAL FILTERS
Milorad Paskaš, Inovation Center, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
Marijeta Savković Ilić, Inovation Center, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
Ana Gavrovska, Inovation Center, School of Electrical Engineering, University
of Belgrade, Belgrade, Serbia
Milan Milivojević, Inovation Center, School of Electrical Engineering,
University of Belgrade, Belgrade, Serbia
Dubravka Jevtić, Inovation Center, School of Electrical Engineering,
University of Belgrade, Belgrade, Serbia
Dragi Dujković, School of Electrical Engineering, University of
Belgrade, Belgrade, Serbia
Irini Reljin, School of Electrical Engineering, University of Belgrade,
Belgrade, Serbia.
In this paper matched filters with crystal units are
simulated on computer. The transfer function is further analyzed and the
influence of the filter parameters on its transfer function in pass-band is
investigated. Simulation of crystal filters enables filter design corrections
before the manufacturing process.
EKI2.2
TOWARDS OPTIMAL CLASSIFIER OF SPECTROSCOPY DATA
David D. Pokrajac, The Optical Science Center for Applied
Research, Delaware State University, Dover, DE, USA,
Poopalasingam Sivakumar, The Optical Science Center for Applied Research,
Delaware State University, Dover, DE, USA,
Yuri Markushin, The Optical Science Center for Applied Research, Delaware
State University, Dover, DE, USA,
Daniela Milovic, Electrical Engineering, University of Nis, Nis, Serbia,
Mukti Rana, The Optical Science Center for Applied Research, Delaware State
University, Dover, DE, USA
Gary Holness, The Optical Science Center for Applied
Research, Delaware State University, Dover, DE, USA
Jinjie Liu, The Optical Science Center for Applied Research,
Delaware State University, Dover, DE, USA
Noureddine Melikechi, The Optical Science Center for Applied Research,
Delaware State University, Dover, DE, USA
Towards Laser spectroscopy can produce vast amounts of
data, anticipating needs for automatization of tasks such as classification
and discrimination of spectra. Using the apparatus of statistical theory of
detection, we develop the optimal classifier for spectroscopy data for a
linear model of an echelle spectrograph system. We validate model assumptions
through statistical analysis of “dark signal” and laser-breakdown induced
spectra of standardized NIST glass. The experimental results suggest that the
quadratic classifier may provide optimal performance if the spectroscopy
signal and noise can be considered Gaussian.
EKI2.3
AN EFFICIENT MATLAB IMPLEMENTATION OF OFDM/OQAM MODULATOR
Selena Vukotić, Faculty of Computer Science, Belgrade,
Serbia
Desimir Vučić, Faculty of Computer Science, Belgrade, Serbia
In this paper we made a theoretical overview considering
OFDM/OQAM signals in continuous and discrete time and presented several
implementation schemes, which were obtained by introducing the filterbank
theory and polyphase decomposition of a prototype filter. Based on the most
efficient scheme, among those presented here, we gave our implementations of
OFDM/OQAM modulator in MATLAB. Based on the simulations results, power
spectrums for different number of subcarriers and pulse shaping parameters are
shown.
ELECTRONICS
ELI1 Electronics Circuits and Systems
Chairs: Tom Kazmierski, University of Southampton, UK
Miroslav Lazić, Iritel a.d. Belgrade, Serbia
Tuesday, June 3, Room 5, 16:00
ELI1.1
OPTIMIZING USE OF THE ALTERNATIVE ENERGY SOURCES - DC
DISTRIBUTION
Invited paper
Miroslav Lazić, Iritel a.d. Belgrade, Serbia
Dragana Petrović, Iritel a.d. Belgrade, Serbia
Zoran Cvejić, Iritel a.d. Belgrade, Serbia
Bojana Jovanović, Iritel a.d. Belgrade, Serbia
Fossil fuels are the basis for electricity production. However, the amount of fossil fuel is
limited. In order to reduce the consumption of fossil fuels, different
solutions of using alternative energy sources such as solar, wind, water and
fuel cells have been realized. Usually, in this solutions, alternative
sources are used separately, or as part of a hybrid system linked to the
power distribution network. In practice, alternative sources are becoming the
elements of distribution system, contributing to the overall production of
electricity. That solution is simple,
but has a low benefit. Problem with this approach is how to count and rely on
the unstable, limited, time variant and unpredictable alternative energy
resources. The electricity distribution network with significant alternative
energy share will strongly depend of the unmanageable natural environment
conditions. Another approach is distributed production, transfer and
consumption of the alternative energy sources. Only a certain part of the energy,
under contract agreement, could be delivered to the power grid network. The
concept proposed here allows more efficient using of alternative energy
sources. It opens many new different possibilities. One of them is DC
electrical power distribution, what will build completely new foundation for
the electrical power system organization. Here we are presenting a set of
devices developed and manufactured as a components for the integrated electrical system build, combining
alternative and traditional energy sources in parallel working.
ELI1.2
HARDWARE-EFFICIENT DIGITAL SYSTEM SYNTHESIS USING A
VARIABLE-ARCHITECTURE APPLICATION-SPECIFIC PROCESSOR
Tom J Kazmierski, Charles Leech and Dominic Murphy,
University of Southampton, United Kingdom
A key design challenge in embedded and mobile systems is to
reduce energy consumption while delivering high performance.
Application-driven specification of a variable-architecture processor is a
promising approach that can produce significant energy saving whilst maintain
or even increase performance. The paper presents a methodology to create an
energy-efficient variable-architecture processor that is tailored to the
specific requirements of an application. As an example, an implementation of
the JPEG decompression algorithm is demonstrated where the hardware size is
reduced four times compared with an equivalent dedicated hardware synthesis
approach.
ELI1.3
ENERGY EFFICIENT CMOS DIGITAL MULTIPLAYERS
Branko L. Dokić, Faculty of Electrical Engineering,
University of Banja Luka, Bosnia and Herzegovina
Velibor Škobić, Faculty of Electrical Engineering,
University of Banja Luka, Bosnia and Herzegovina
Aleksandar Pajkanović, Faculty of Technical Sciences,
University of Novi Sad, Serbia
In this paper the characteristics of CMOS digital array
multipliers in three operating regimes (weak, strong and mixed inversion) are
given. Two topologies characterized by series transmission – dual and
symmetric transistor networks – are used. It is shown that the mixed CMOS
regime is the optimum choice in the context of low consumption and high
speed. The characteristics compared are obtained using PSPICE 16.3 and the
parameters of a 180 nm technology.
ELI1.4
HIGH ACCURANCY SELF-CONFIGURABLE DLL BY FREQUENCY RANGE
Vazgen Melikyan, Synopsys Armenia Educational Department,
Yerevan, Armenia
Arthur Sahakyan, Synopsys Armenia Educational Department,
Yerevan, Armenia
Mikayel Piloyan, Synopsys Armenia Educational Department,
Yerevan, Armenia
Armen Sahakyan, State Engineering University of Armenia, Yerevan,
Armenia
Borisav Jovanović, Faculty of Electronic Engineering,
University of Niš, Serbia
An architecture of Self-configurable Delay-Locked Loop
(SCDLL) by frequency range presented in this paper. The proposed architecture
produces different phases of clock signal for the different frequencies in
the output of DLL over PVT, which is needed to avoid clock skew and
setup/hold time margins violations between the different synchronized blocks
inputs. The presented architecture is self-configurable depended on input
clock signal frequencies and operates by the negative feedback system. DLL is
widely used in such circuits where is the actual issue to get clock signals
with different phases. For example in special input/output circuits of
several standards such as Universal Serial Bus (USB), Double Data Rate (DDR)
and etc.
ELI1.5
NEW BLOOD LEVEL DETECTION SYSTEM IN BLOOD SEPARATING MACHINE
Miloš Petković, Faculty of Electronic Engineering,
University of Niš, Serbia
Miroslav Božić, Faculty of Electronic Engineering,
University of Niš, Serbia
Dragiša Krstić, Faculty of Electronic Engineering,
University of Niš, Serbia
Darko Todorović, Faculty of Electronic Engineering,
University of Niš, Serbia
Goran S. Đorđević, Faculty of Electronic
Engineering, University of Niš, Serbia
Standard version of a blood separators typically use
medium-price color sensors for detection of boundary level between red blood
cells and plasma, at the last gate – at hose clamps. Discrete number of
sensors are related to a number of significant levels to be detected thus
making blood separation potentially faulty and unreliable. Our target was to
make flexible, low cost replacement for level detection system that can be
easily integrated into the existing product. We came up with an image
processing solution that uses USB web-camera, ARM based off-the-shelf board –
BeagleBone black and free OpenCV library. Flexibility is held in much higher,
selectable number of levels, freely positioned USB camera and brand-free independent
processing platform, as well as semi-automatic calibration system. By adding
minimum additional electronics, we managed to integrate our solution into
existing Blood processing machine. In conclusion, we added new value to the
machine at lower cost in production, increasing measurement frequency and
resolution needed for improvement of blood separation process. Next step is
to try to use two USB cameras on a custom-made board, for simultaneous level
detection on two channel blood separator, bringing the system integration to
the higher level.
ELI1.6
RACPBERY PI AS INTERNET OF THINGS HARDWARE: PERFORMANCES AND
CONSTRAINTS
Mirjana Maksimović, Faculty of Electrical Engineering,
University of East Sarajevo, Bosnia and Herzegovina
Vladimir Vujović, Faculty of Electrical Engineering,
University of East Sarajevo, Bosnia and Herzegovina
Nikola Davidović, Faculty of Electrical Engineering,
University of East Sarajevo, Bosnia and Herzegovina
Vladimir Milošević, Faculty of Technical Sciences,
University of Novi Sad, Serbia
Branko Perišić, Faculty of Technical Sciences,
University of Novi Sad, Serbia
The Internet of Things (IoT) ideology can be looked as a
highly dynamic and radically distributed networked system composed of a very
large number of identifiable smart objects. These objects are able to
communicate and to interact among themselves, with end-users or other
entities in the network. Entering the era of Internet of Things, the use of
small, cheap and flexible computer hardware that allow end-user programming
become present. One of them, considered in this paper, is the Raspberry Pi,
fully customizable and programmable small computer board. Comparative
analysis of its key elements and performances with some of current existing
IoT prototype platforms have shown that despite few disadvantages, the
Raspberry Pi remains an inexpensive computer with its very successfully usage
in diverse range of research applications in IoT vision
ELI1.7
ON THE METHOD DEVELOPMENT FOR ELECTRICITY LOAD FORECASTING
Jelena Milojković, Innovation centre of advanced
technologies, Niš, Serbia
Vančo Litovski, Faculty of Electronic Engineering,
University of Niš, Serbia
The way of development a method for systematic prediction
of electricity load at suburban level is described. Steps that were passed
are described and the evolution of the method is illustrated by proper
examples. Prediction of electricity load at annual, monthly, daily and hourly
level as well as peak consumption is considered. Comparisons with some other’s
results are given, too.
ELI1.8
EXPERIENCE IN USING OPEN COMMAND ENVIRONMENT FOR ANALYSIS IN
EDUCATION
Dejan Mirković, Faculty of Electronic Engineering,
University of Niš, Serbia
Predrag Petković, Faculty of Electronic Engineering,
University of Niš, Serbia
This paper sublimates experiences in using Open Command
Environment for Analysis (OCEAN) for teaching IC design at University of Niš.
IC design requires a lot of repetitive analysis needed to get better insight
into possibilities of a new or unknown process. Since design automation
implies usage of appropriate software environment, one such environment known
as OCEAN will be considered. Basic guidelines and techniques for exploiting
it will be given. Concrete example will demonstrate usefulness of the platform.
BIOMEDICINE
MEI.1 Biomedicine
Chair: Dejan Popović, School of Electrical
Engineering, University of Belgrade, Serbia
Tuesday, June 3, Room 5, 14:00
MEI1.1
ASSESMENT OF VIGILANCE BASED ON EEG SIGNALS FOR PREDICTION OF
THE SPEAKER/LISTENER INTERACTION
Milena Okošanović (Student) mBrainTrain d.o.o., Faculty
of Electrical Engineering, University of Belgrade, Serbia
Ivan Gligorijević, mBrainTrain d.o.o., Faculty of
Engineering, University of Kragujevac, Serbia
The objective of the present work in progress is to set up
a novel approach for on-line tracking of psychophysiological markers for
detection of slips in attention and vigilance decrement in an educational
environment. We present here the recordings and processing method that allows
the estimation of the level of vigilance from the cortical signals measured
at the scull and when using the Smarting® device (24 channels digital
amplifier with wireless communication with a Windows/Android platform).
Wireless electroencephalographic measurements (EEG) were used for the
assessment of subject's concentration during a lecture and estimation of the
optimum duration of the lecture. The optimum is defined as the longest time
during which the vigilance decrement raises over the predefined threshold.
The state when the vigilance drops below the specific level can be considered
as the state of drowsiness. This method allows also the training of the
teacher since the feedback from the listener can be brought to the teacher
on-line.
MEI1.2
HEURISTIC SYNTHESIS OF JOINT ANGLES BASED ON DATA FROM
INERTIAL SENSORS DURING GAIT
Marija Petrović (Student), University of Belgrade,
School of Electrical Engineering, Serbia
Dejan Popović,
University of Belgrade, School of Electrical Engineering, Serbia
The gait performance is of interest for the diagnostics and
the assessment of the efficacy of a therapy in rehabilitation. The
availability, prices, accuracy, reproducibility and ease of calibration of
body worn inertial sensors make the excellent candidates for the clinical
assessment of the gait performance. The output data from inertial sensors
needs to be processed in order to generate joint angles and other data in a
form suitable for clinicians. We suggest the application of principal
component analysis (PCA) for the recognition of gait modality, and a
heuristic fitting of predefined joint angle patterns between the
automatically recognized gait events. The heuristic fitting aims to generate the angle joint
data that is matching the data recorded with joint angle sensors. We tested
this method for normal and artificially made abnormal gait (walking on toes,
walking on heels, etc.) in healthy subjects. The application of this method
and instrumentation is envisoned in the rehabilitation of humans with
impaired gait after stroke, surgery, or other disease affecting bipedal
progression.
MEI1.3
BIOFEEDBACK: SOFTWARE FOR PROCESSING OF MULTIPLE
ELECTROPHYSIOLOGICAL SIGNALS
Ilija Jovanov (Student), University of Belgrade, School of Electrical
Engineering, Serbia
Dejan Popović, University of Belgrade, School of
Electrical Engineering, Serbia
Biofeedback is the process of gaining greater awareness of
a physiological function with the aim of improving the performance. This
improvement is due to cortical plasticity; thus, biofeedback could be
considered as a "training of the brain". We show here the software
that was developed for the Smarting® device, developed by the Serbian company
"mBrainTrain". Smarting is a light body-worn 24-channel recording
interface of small signals (>1 µV) and Bluetooth transmission of digital
outputs to Windows/Android platform. Communication, processing, and control,
are handled by software developed in C++, while LabVIEW software is used for
graphical representation of signals to the user (biofeedback). This system
can be used for assessing muscle activity by surface electrodes (sEMG), brain activity with the appropriate
electrode sets (EEG, evoked potentials), or other information coming from the
body worn sensors (e.g., ECG, arousal - GSR, EOG). The use is primarily
envisioned in the domain of neurorehabilitation of humans with brain injury
or disease.
MEI1.4
EEG SIGNAL PROCESSING AND CLASSIFICATION USING ANN
Damir J. Đozić (Student), University of Novi Sad,
Faculty of Technical Sciences, Serbia
Goran Krajoski (Student), University of Novi Sad, Faculty of
Technical Sciences, Serbia
Nikola Popov (Student), University of Novi Sad, Faculty of
Technical Sciences, Serbia
Darko Stanišić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Slađana Lazarević (Student), University of Novi
Sad, Faculty of Technical Sciences, Serbia
There is a wide area of application for good interpretation
of Electroencephalogram (EEG) signals, from medicine to computer games. In
order to be used for control, EEG signals must be acquired and classified in
the right way. This research includes extraction of 15 new features from EEG
signals in frequency domain (not yet seen in literature), and classification
of signals using those features. Classification is done with Artificial
Neural Network method, and the results are found to be promising.
MEI1.5
LABORATORY SYSTEM FOR PHYSIOLOGICAL MEASUREMENTS
Nikola Jorgovanović, University of Novi Sad, Faculty of Technical
Sciences, Serbia
Mišo Stanković (Student), University of Novi Sad,
Faculty of Technical Sciences, Serbia
Željko Tepić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Vojin Ilić, University of Novi Sad, Faculty of Technical
Sciences, Serbia
Dubravka Bojanić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
This paper presents the design of a physiological
measurements system, which allows recording of electrophysiological signals
(electrocardiography ECG, electromyography EMG and electroencephalography
EEG). The system also has the ability to monitor the concentration of oxygen
in arterial blood using the method of pulse oximetry. Its main purpose is the
use for educational purposes, but can also be used as a development system
for the devices that are based on some of the applied integrated
circuits.
METROLOGY
MLI1. MEASUREMENT METHODS
Chair: Platon Sovilj, Faculty of Technical Sciences,
University of Novi Sad, Serbia
Monday, June 2, Room 3, 14:00
MLI1.1
DENSITY MEASUREMENT USING THE OSCILLATORY TYPE DEVICES
Emil Peić Tukuljac, School of Professional Higher
Education, Subotica, Serbia
Zoran Mitrović, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Nemanja Gazivoda, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Božidar Vujičić, University of Novi Sad, Faculty of
Sciences, Serbia
Jovan Mitrović (Student), University of Novi Sad,
Faculty of Technical Sciences, Serbia
Density of fluids
may be determined using different measuring devices. Recently, a wide variety
of oscillatory devices are used to measure the density. Their application
allows sophisticated analysis of fluids in biotechnology, pharmaceutics,
saturation processes, concentration or crystallization procedures. Unique
design allows temperature compensation and obtaining the correct measurements
in various measurement processes. In the paper, the mathematical model and
the potential uncertainty in the calibration of the aforementioned devices
will be described.
MLI1.2
IDENTIFICATION OF THE MODEL OF FREQUENCY VARIATIONS IN POWER
GRID
Željko Beljić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Duško Davidović, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Bojan Vujičić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Dragan Pejić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Vladimir Vujičić, University of Novi Sad, Faculty
of Technical Sciences, Serbia
This paper presents the analysis of the model of frequency
variations in Serbia, as well as in the European interconnection. Multiple
measurements have shown that the distribution of frequencies in Serbia, to
the greatest extent, is normal - Gaussian. It is known that the frequency
variations affect the measurement of the power grid parameters, and on
example of the mains voltage THD factor measurement the impact of frequency
variations to measurement error is shown. Results obtained from a series of
simulations are presented in the paper. Those results represent a significant
contribution to a better understanding of the power grid behavior, as well as
to the further development of measurement methods for measurement of the
power grid parameters.
MLI1.3
STOCHASTIC MEASUREMENT IN FOURIER AND WAVELET DOMAIN: A
COMPARATIVE STUDY
Aleksandar Radonjić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Platon Sovilj, University of Novi Sad, Faculty of Technical
Sciences, Serbia
Vladimir Vujičić, University of Novi Sad, Faculty
of Technical Sciences, Serbia
In this paper we compare two strategies for measuring in
frequency domain using digital stochastic measurement (DSM) method. One of
them, based on Fourier series expansion, was already analyzed in theory and
applied in practice. The second strategy, based on Wavelet series expansion,
has never been discussed in literature.
MLI1.4
STOCHASTIC MEASUREMENT OF THE SPECTRAL POWER DENSITY OF
LOW-FREQUENCY NOISE GENERATED BY GRAPHENE-BASED CHEMICAL AND BIOLOGICAL SENSORS
– CASE STUDY
Vladimir Vujičić, University of Novi Sad, Faculty
of Technical Sciences, Serbia
Aleksandar Radonjić,
University of Novi Sad, Faculty of Technical Sciences, Serbia
Bojan Vujčić, University of Novi Sad, Faculty of Technical
Sciences, Serbia
Željko Beljić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
This paper proposes a new method for measuring the spectral
power density of low-frequency noise generated by graphene-based chemical and
sensors. The proposed solution is based on two-bit digital stochastic
measurements (DSMs). In addition, we criticize the common measurement method
based on application of Fast Fourier Transform (FFT) and digital signal
processors (DSPs). It is shown that this method requires large hardware
resources and great power supply.
MLI1.5
APROXIMATION OF THE NTC RESISTOR CURVE BY NEURAL NETWORK
USING MICROCONTROLLERS
Marko Dragoslavić, School of Electrical Engineering,
University of Belgrade, Serbia
Ivan Božić, School of Electrical Engineering, University
of Belgrade, Serbia
Marija Cvijanović, Faculty of Agriculture, University of
Belgrade, Serbia
Marko Cvijanović, School of Electrical Engineering,
University of Belgrade, Serbia
The objective of this paper is to demonstrate the approximation
of NTC resistor R/T curve using neural networks. Special attention is focused
on the efficiency of the approximating algorithm considering that
microcontrollers have limited memory and processing ability. Dedicated
systems often use battery power to periodically measure temperature and
therefore it is important to carry out approximation of the algorithm in the
least possible number of beats. We compared the algorithm based on neural
network with interpolation of temperature. The paper will show that the
algorithm which uses data from previously trained neural network can be more
efficient than the conventional method of approximation using temperature
polynomial. For the needs of this paper the temperature range between 00C
and 1200C was selected, which was tested on three generations of
microcontrollers.
MLI2. Measurement instruments and systems
Chair: Dragan Denić, Faculty of Electronics
Engineering, University of Niš
Monday, June 2, Room 3, 16:00
MLI2.1
TEMPERATURE AND PRESSURE MEASUREMENT IN VALIDATION PROCEDURE
FOR FOOD PRODUCT PASTEURISATION
Emil Peić Tukuljac, School of Professional Higher
Education, Subotica, Serbia
Nemanja Gazivoda,
University of Novi Sad, Faculty of Technical Sciences, Serbia
Zoran Mitrović, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Ivan Župunski, University of Novi Sad, Faculty of Technical
Sciences, Serbia
Nemanja Čabrilo, University of Novi Sad, Faculty of
Technical Sciences, Serbia
The process of thermal treatment of food products provides
long shelf life while preserving the quality without the use of chemical
additives. Regardless of whether it is a short-term process under heating for
a few seconds, or an ultra high temperature (up to 140 °C) or the heating
method which requires a longer thermal treatment, but at lower temperature
(88-90 °C) it is important to monitor
the critical process parameters. Getting a quality product involves not only thermal treatment of the
contents of the packaging, but also all surfaces that have contact with the
product.
MLI2.2
MONITORING OF THE SEISMIC ACTIVITY USING HIGH SENSITIVITY
MEMS ACCELEROMETER
Djordje Klisić, School of Electrical Engineering,
University of Belgrade, Serbia
Ilija Radovanović, School of Electrical Engineering,
University of Belgrade, Serbia
Nikola Bežanić, School of Electrical Engineering,
University of Belgrade, Serbia
Ljubisav Stamenić, Vinča Institute of Nuclear
Sciences, University of Belgrade, Serbia
Ivan Popović, School of Electrical Engineering,
University of Belgrade, Serbia
MEMS sensors are widely used type of sensors that can be
found in many modern gadgets. Affordable price and small dimensions in
combination with high sensitivity and precision made them interesting
candidates for the wide range applications in various fields. In case of
seismology, application of MEMS accelerometer is still in the shadow of the
classical accelerometers with the inertial element, although the
characteristics of the today MEMS accelerometers can easily match the
classical ones. In this paper group of authors decided to test
characteristics of the high sensitivity accelerometer and to determine if it
is suitable for the applications in seismology. The requirements that sensor
had to meet are high sensitivity and noise level below the threshold of human
detectable vibrations which is around four degrees of the Mercalli scale or
the ground acceleration between 0.015g and 0.04g.
MLI2.3
IMPLEMENTATION OF INDUSTRIAL NETWORKS AND PROTOCOLS IN THE
LabVIEW PROGRAM
Josif Tomić, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Miodrag Kušljević, Termoelektro Enel AD, Belgrade,
Serbia
Milan Vidaković, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Miloš Živanov, University of Novi Sad, Faculty of Technical
Sciences, Serbia
Modern industry can no longer be imagined without the
application of information technologies and complex network systems.
Industrial processes are becoming increasingly complex and require the integration
and synchronization of a great number of the most various measurement
sensors, PLC devices, microprocessors, actuators, IO devices, motors and
other equipment. Networking of this equipment is carried out on several
hierarchical levels where the lowest level usually has requirements for
operation in real time, while the higher levels are not as time-critical and
have lately been relying on Internet protocols. There are several dozens of
different industrial networks and protocols currently functioning on the
market, and knowledge of them represents a very complex task for engineers
that deal with the automation of industrial processes. This paper presents
one management solution that uses Modbus protocol that is based on the
LabVIEW program. The LabVIEW program significantly simplifies this task as a
result of its exceptional flexibility and a large number of pre-written
subroutines for a large number of industrial networks and protocols.
Furthermore, the LabVIEW program also contains some original solutions that
significantly simplify the implementation of complex SCADA systems, at the
same time allowing them to operate in real time.
MLI2.4
STOCHASTIC CONTROL OF PARALLEL CONNECTED DC/DC CONVERTERS
Velibor Pjevalica, JP Srbijagas Novi Sad, Serbia
Nebojša Pjevalica, University of Novi Sad, Faculty of
Technical Sciences, Serbia
Vladimir Vujičić, University of Novi Sad, Faculty
of Technical Sciences, Serbia
In order to
process higher and higher power, DC-DC converters should be operating in
parallel. Parallel processing of power means higher efficiency and better
reliability. In this paper, stochastic control for parallel DC-DC converters
structures is presented.
MLI2.5
THE THEOREM ABOUT THE TRANSFORMER EXCITATION CURRENT WAVEFORM
MAPPING INTO THE DYNAMIC HYSTERESIS LOOP BRANCH FOR THE SINUSOIDAL MAGNETIC
FLUX CASE
Nenad Petrović, Electrotechnical school Stari grad,
Belgrade, Serbia
Velibor Pjevalica, JP Srbijagas Novi Sad, Serbia
Vladimir Vujičić, University of Novi Sad, Faculty
of Technical Sciences, Serbia
The paper
analyses aspects of the approximation theory usage on the certain subsets of
the measured samples of the excitation transformer current and the sinusoidal
magnetic flux in the single-phase transformer, Epstein frame and toroidal
core specimen. In the paper is given the theorem of direct mapping the
excitation transformer current in the stationary regime to the dynamic
hysteresis loop branch (in further text DHLB). This theorem provides the
necessary and satisfactory conditions for above mentioned mapping. The
theorem highlights that the excitation transformer current under the
sinusoidal magnetic flux has qualitatively equivalent information about
magnetic core properties as the DHLB. Furthermore, the theorem establishes
direct relationship between the number of the excitation transformer current
harmonics and their coefficients with the degree and the coefficients of the
DHLB interpolation polynomial for the subsets of the measured samples that
represent the Chebyshev nodes of the first and the second kind. These
nonequidistant Chebyshev fi-i0 nodes provides the uniform convergence of the
interpolation polynomial to the experimentally obtained DHLB with an
excellent approximation accuracy and are applicable on the approximation of
the static hysteresis loop branches and the DC magnetization curves as well.
MLI2.6
STUDY OF A ROOM TEMPERATURE AND LIGHT DYNAMIC
Razvan Craciunescu, Telecommunication Department, Faculty of
Electronics, Telecommunications and Information Tehnology, Bd. Iuliu Maniu, 1-3, 061071, Bucharest,
Romania
Andreea Garliceanu, Telecommunication Department, Faculty of
Electronics, Telecommunications and Information Tehnology, Bd. Iuliu Maniu, 1-3, 061071, Bucharest,
Romania
Octavian Fratu, Telecommunication Department, Faculty of
Electronics, Telecommunications and Information Tehnology, Bd. Iuliu Maniu, 1-3, 061071, Bucharest,
Romania
ZigBee wireless
technology makes the control network applications and the wireless sensors
more practical. The usage of this technology
can offer security, light control, access control to your home. Building
Automation is a very wide market nowadays. Smart Energy is a big part of
Building Automation. Some of the features included in the Smart Energy
profile are: basic metering, pricing, support for water and gas, security to
allow consumer only, text messaging in case of incidents. In this paper we use a ZigBee based
wireless sensor network in order to study the dynamic of light and temperature
in a room. The measurement were done in different environment conditions.
MICROELECTRONICS AND OPTOELECTRONICS
MOI1.
Microelectronics and MEMS
Chair: Zoran
Prijić, University of Niš, Serbia
Monday, June 2, Room 5, 14:30
MOI1.1
WET ISOTROPIC CHEMICAL ETCHING OF PYREX GLASS WITH Cr/Au
MASKING LAYERS
Žarko Lazić, Centre
of Microelectronic Technologies, Institute of Chemistry, Technology and
Metallurgy, University of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
Milče M.
Smiljanić, Centre of Microelectronic Technologies, Institute of
Chemistry, Technology and Metallurgy, University of Belgrade, Njegoševa 12,
11000 Belgrade, Serbia
Milena Rašljić,
Centre of Microelectronic Technologies, Institute of Chemistry,
Technology and Metallurgy, University of Belgrade, Njegoševa 12, 11000
Belgrade, Serbia
Ivana
Mladenović, Centre of Microelectronic Technologies, Institute of
Chemistry, Technology and Metallurgy, University of Belgrade, Njegoševa 12,
11000 Belgrade, Serbia
Katarina
Radulović, Centre of Microelectronic Technologies, Institute of
Chemistry, Technology and Metallurgy, University of Belgrade, Njegoševa 12,
11000 Belgrade, Serbia
Milija
Sarajlić, Centre of Microelectronic Technologies, Institute of
Chemistry, Technology and Metallurgy, University of Belgrade, Njegoševa 12,
11000 Belgrade, Serbia
Dana
Vasiljević-Radović, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
In this paper we developed various techniques of wet
isotropic chemical etching of Pyrex glass in an aqueous solution of 49% HF.
The techniques are based on the processes of sputtering and photolithography.
The various thin films of Cr/Au were sputtered. Layers of Cr/Au and
photoresist serve as a masking material during etching of Pyrex glass in
aqueous HF.
MOI1.2
IMPLEMENTATION OF IEEE 1451.4 TRANSDUCER ELECTRONIC DATA
SHEET (TEDS) USING AN 'MBED' MICROCONTROLLER MODULE: THE CASE OF MEMS
PIEZORESISTIVE PRESSURE SENSORS
Branko Vukelić, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
Miloš Frantlović, Centre of Microelectronic
Technologies, Institute of Chemistry, Technology and Metallurgy, University
of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
In this paper we describe an implementation of the IEEE
1451.4 standard, based on the "mbed" hardware and software platform
which enables simple and efficient development of microcontroller-based
solutions. The Class 2 Mixed Mode Interface (MMI), defined by the standard,
was chosen for IHTM silicon piezoresistive MEMS pressure sensors used in
intelligent pressure transmitters. A program for the microcontroller is
written, enabling communication with the EEPROM memory containing the
Transducer Electronic Data Sheet (TEDS), as well as with a personal computer
via the USB virtual serial port. A user application is also created for
communication with the "mbed" module, enabling user-friendly
management of the TEDS contents.
MOI2. nanoETRAN
Chair: Zoran Jakšić, ICTM, University of Belgrade, Serbia
Monday, June 2, Room 5, 16:00
MOI2.1
PAULI PARAMAGNETISM AND SPIN-VALLEY FILTERING IN GRAPHENE VAN
DER WAALS HETEROSTRUCTURES
M. Grujić, School of Electrical Engineering, University
of Belgrade, Belgrade, Serbia and Department of Physics, University of
Antwerp, Antwerp, Belgium
M. Ž. Tadić, School of Electrical Engineering,
University of Belgrade, Belgrade, Serbia
F. M. Peeters, Department of Physics, University of Antwerp,
Antwerp, Belgium
The most recent
trend in 2D materials research is reflected in the idea of stacking different
material layers on top of each other, forming the so called van der Waals
heterostructures, in order to custom tailor the device properties according
to specific needs. In the light of this paradigm shift, and for the
particular case of graphene, tantalizing new experiments reveal that carrier
mass and spin-orbit interaction, properties that are vanishing in intrinsic
graphene, can be artificially engineered. The interplay of the two parameters
is especially interesting given that they open topologically distinct gaps.
Given that the additional valley degree of freedom arises in graphene due to
its honeycomb structure, the aforementioned interplay will also result in a
spin and valley dependent gap in bulk graphene. Here we show that this
will influence the transmission through a 1D barrier made of this two
parameters in the presence of a magnetic field significantly. Specifically,
the two parameters induce Pauli paramagnetic moments which manifest in the
transmission of the electron through the barrier. We find that these magnetic moments couple with
spin-orbit interaction and with mass through spin and valley degrees of
freedom, which is reflected in the Landau level spectrum.
Additionally, we show that only opposite spins from opposite valleys can be
transmitted given a proper set of parameters. On the basis of these findings
we demonstrate that in the presence of a strain-induced pseudomagnetic field,
and due to the finite mass and spin-orbit interaction, a spin-valley
filtering device can be obtained. We also show that the filtering behavior is
controllable by electrical gating as well as by strain.
MOI2.2
CONDUCTION-BAND NONPARABOLICITY AND GAIN CALCULATIONS FOR THz
QUANTUM CASCADE LASER IN STRONG MAGNETIC FIELD
Nikola
Vuković, School of Electrical Engineering, University of
Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Aleksandar Daničić, Vinca Institute of
Nuclear Sciences, University of Belgrade, 11001 Belgrade, Serbia
Jelena
Radovanović, School of Electrical Engineering, University of
Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Vitomir Milanović,
School of Electrical Engineering, University of Belgrade, 73 Bulevar
kralja Aleksandra, 11020 Belgrade, Serbia
Simulation results for the optical gain and scattering
rates of relevant transitions in the THz quantum cascade laser (QCL)
structures subjected to strong magnetic field are presented. Starting from a
comprehensive nonparabolic model of III-V semiconductor conduction-band we
first obtain bound states energies of a QCL without the presence of magnetic
field. A strong magnetic field applied along the growth direction of the
structure provides discretization of the energy spectrum into Landau levels
(LLs). Once familiar with the electronic structure, we are able to evaluate
longitudinal optical (LO) phonon and interface roughness (IR) scattering
rates and use them as input data for the system of rate equations. After we
calculate the population inversion from this system, we can estimate the
optical gain at desired wavelength.
MOI2.3
TAILORABLE PLASMONIC RESPONSE OF FREESTANDING METAL-COMPOSITE
NANOMEMBRANES WITH 2D ARRAYS OF SUBWAVELENGTH CIRCULAR APERTURES
Zoran Jakšić, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
Jovan Matović, Department for Energetics, Electronics
and Telecommunications, Faculty of Technical Sciences, University of Novi
Sad, Novi Sad, Serbia
Alois Lugstein, Institut für Festkörperelektronik, Technische
Universität Wien, Floragasse 7, 1040 Wien, Österreich
Marko Obradov, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
We report ab initio
electromagnetic modeling and
nanofabrication of 7 nm thick metal-containing freestanding nanomembranes
with square lattice of circular apertures (single fishnet structure) 100 nm
in diameter, intended for plasmonic applications. We used finite element
modeling to determine the scattering parameters of our nanomembranes for different
plasmonic materials (silver, chromium). The nanomembrane samples were
fabricated using radiofrequent sputtering of chromium to single crystalline
silicon wafer, while bulk micromachining in KOH was used to etch away silicon
support (sacrificial structure). The nanoaperture arrays (square lattice with
a 300 nm lattice constant) were fabricated directly in the freestanding
metal-composite nanomembrane using the focused ion beam technique. Large
absorption in visible has been observed both theoretically and experimentally
that can be contributed to impedance matching between the nanomembrane and
free space. The final result is a perfectly symmetrical platform for long
range surface plasmon polaritons with a possibility to tailor its response
utilizing designer plasmons. The choice of geometrical and material
parameters allows for tuning of frequency dispersion and effective parameters
of the structure and results in rich spectral behavior in spite of the
apparent simplicity of the structure.
MOI2.4
INFRARED PHOTODETECTOR ENHANCEMENT UTILIZING TRANSPARENT
CONDUCTIVE OXIDE SUBMICROMETER PARTICLES EMBEDDED IN GRADIENT INDEX
ANTIREFLECTION LAYER
Marko Obradov, Centre
of Microelectronic Technologies, Institute of Chemistry, Technology and
Metallurgy, University of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
Zoran Jakšić, Centre
of Microelectronic Technologies, Institute of Chemistry, Technology and
Metallurgy, University of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
Dana
Vasiljević-Radović, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
We consider the enhancement of the specific detectivity of
semiconductor infrared photodetectors for mid-infrared range utilizing
localized surface plasmons resonance (LSPR) of spherical submicrometer
plasmonic particles. A plasmonic particle localizes electromagnetic field
within the detector active area, but conventional plasmonic materials, like
noble metals, have their resonant frequencies in the ultraviolet or visible
part of the spectrum. In this work we investigate a redshifting strategy
based on simultaneous use of doped spherical submicrometer transparent
conductive oxide (TCO) plasmonic particles and an embedding high-permittivity
dielectric that further redshifts the spectral characteristics. In order to
avoid large reflection losses at the surface of embedding dielectric, we
apply on the top of it an antireflective layer with a linear increase of
refractive index. We perform an ab
initio simulation of the whole system starting from Maxwell equations and
utilizing the finite element method. We determine the scattering properties
of the system including scattering cross sections and spatial electromagnetic
field distribution. We confirm a resonant response of TCO particles in
mid-infrared, tailorable by adjusting the surrounding medium permittivity.
The use of the antireflective layer is crucial for achieving scattering
cross-sections more than 20 times larger than the geometric cross-section.
MOI2.5
SELECTIVITY ISSUES IN AFFINITY-BASED BIOCHEMICAL SENSORS:
DETERMINING THE RATIO OF SIMILAR BIOMOLECULES IN BINARY MIXTURES
Ivana Jokić, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
Olga Jakšić, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
Zoran Jakšić, Centre of Microelectronic Technologies,
Institute of Chemistry, Technology and Metallurgy, University of Belgrade,
Njegoševa 12, 11000 Belgrade, Serbia
Katarina Radulović, Centre of Microelectronic
Technologies, Institute of Chemistry, Technology and Metallurgy, University
of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
Miloš Frantlović, Centre of Microelectronic
Technologies, Institute of Chemistry, Technology and Metallurgy, University
of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
We consider selectivity of affinity-based nanosensors
utilizing resonance shift due to the presence of adsorbed analyte. Among such
devices we analyze mass-based sensors utilizing mechanical resonance in e.g.
micro or nanocantilevers and all-optical refractometric sensors utilizing
surface plasmon polariton resonance. The sensitivity of such devices can be
extremely high, reaching single-molecule level, however their selectivity is
limited by the differences in mass in the first case or in refractive index
values between different analytes in the second. The typical approach is to
use some kind of receptors on the sensor surface with highly specific binding
of a targeted analyte. The properties of a given biomolecule, for instance
protein, will vary between its different conformations due to different
arrangements of their atoms in space. Since the conformation of a molecule is
critically important for its function, it is of interest to determine the
ratio between different conformational isomeres in a given sample. In
general, different conformations of a biomolecule may have different affinity
toward binding sites on the surface of an affinity-based nanosensor, as well
as different surface-volume ratios. We argue that the analysis of
adsorption kinetics ensures sufficient
data to discriminate between different conformations in a mixture.
MOI2.6
UNIT-CELL LEVEL
SUPERSTRUCTURES FOR THE EXTENSION OF SPECTRAL RANGE OF DOUBLE FISHNET
METAMATERIAL PARAMETERS AND FINE TUNING OF THEIR EFFECTIVE OPTICAL PROPERTIES
Dragan Tanasković,
Centre of Microelectronic Technologies, Institute of Chemistry,
Technology and Metallurgy, University of Belgrade, Njegoševa 12, 11000
Belgrade, Serbia
Zoran Jakšić,
Centre of Microelectronic Technologies, Institute of Chemistry,
Technology and Metallurgy, University of Belgrade, Njegoševa 12, 11000
Belgrade, Serbia
Marko Obradov, Centre
of Microelectronic Technologies, Institute of Chemistry, Technology and
Metallurgy, University of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
Olga Jakšić, Centre
of Microelectronic Technologies, Institute of Chemistry, Technology and
Metallurgy, University of Belgrade, Njegoševa 12, 11000 Belgrade, Serbia
Ivana
Mladenović, Centre of Microelectronic Technologies, Institute of
Chemistry, Technology and Metallurgy, University of Belgrade, Njegoševa 12,
11000 Belgrade, Serbia
We considered a possible generalization of aperture-based
metamaterial structures for the optical range like single and double fishnets
where a single aperture is a compound object formed by superposition of two
or more primitives (rectangles, circles, ellipses). Thus obtained "super
unit cell" reflects the properties of all subobjects that comprise it.
The newly introduced features may be utilized to fine tune the effective
optical parameters of the structure like its scattering parameters and thus
the effective permittivity, permeability and refractive index. They can also
be used to extend and modify in other ways the spectral characteristics of
the structure, thus resulting in enhanced modes of operation. In our
investigation we applied finite element modeling and analytical calculation
to the case of 2D array of square apertures where an aditional smaller square
aperture was placed in each corner of the fundamental aperture. It is our
conclusion that the obtained compound super unit cells can find application
in different fields like waveguiding and sensing.
MOI3.
Optoelectronics
Chair: Ljubiša Tomić, Technical Test Center, Serbia
Tuesday, June 3, Room 5, 8:30
MOI3.1
NUMERICAL ESTIMATION OF THE NOISE EQUIVALENT TEMPERATURE
DIFFERENCE OF THE THIRD GENERATION THERMAL IMAGERS
Anis Redjimi, Military Academy, University of Defense,
Belgrade, Serbia
Nikola
Jovanović, Military Technical Institute, Ratka
Resanovića 1, 11000 Belgrade, Serbia
Dragan Knežević, Military Technical Institute, Ratka
Resanovića 1, 11000 Belgrade, Serbia
Ljubiša Tomić, Technical Test Center, Vojvode
Stepe 445, 11020 Beograd, Serbia
This paper proposes a software tool for numerical estimation
of noise equivalent temperature difference (NETD) which allows the user to
see and check the effect of each thermal imager parameter. The developed
software is based on a mathematical model designed to estimate the NETD for
1st and 2nd generation thermal imagers which is modified to estimate the NETD
for 3rd generation thermal imagers. As a demonstration, two models of thermal
imagers of the third generation are tested. The obtained results are compared
with the declared values given by the producers. Moreover, the developed
software tool could be used as an educational tool for examination and
demonstration of thermal imager properties.
MOI3.2
NUMERICAL SIMULATION OF THE BLACK BODY RADIATION LAWS
Katarina Savić, Military Technical
Institute, Ratka Resanovića 1, 11000 Belgrade, Serbia
Anis Redjimi, Military
Academy, University of Defense, Belgrade, Serbia
Nikola Jovanović, Military
Technical Institute, Ratka Resanovića 1, 11000 Belgrade, Serbia
Dragan Knežević, Military Technical Institute, Ratka Resanovića 1, 11000 Belgrade,
Serbia
This paper proposes a software tool for numerical
simulation of black body radiation laws which allows the user to see and
check the effect of each parameter that influences the global curve of radiation.
It can also be used to calculate each parameter separately knowing the other
necessary parameters. The developed software could be used as an educational
tool for examination and demonstration of basic principles of thermography
and infrared radiation.
MICROWAVE TECHNIQUE, TECHNOLOGIES AND SYSTEMS
Session MTI1
Chairs:
Bratislav Milovanović, University
of Niš, Serbia
Nebojša
Dončov, University of Niš, Serbia
Tuesday, June 3, Room 4, 8:30
MTI1.1
PRINTED
ANTENNA ARRAY WITH TAPERED DIPOLES IMPEDANCES
Marija
Milijić, Faculty of
Electronic Engineering, University of Niš, Niš, Serbia
Aleksandar
Nešić, IMTEL
Komunikacije, Novi Beograd, Serbia
Bratislav
Milovanović, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
This paper
presents printed antenna array which consists of symmetrical dipoles and
symmetrical in-phase feeding line. Dolph-Chebyshev distribution has been used
enabling side lobe suppression around 30 dB in E-plane in the range of
approximately 10 GHz. Dipoles, operating at a second resonance, have
different impedances that are calculated in order to their powers match
specified distribution enabling great side lobe suppression. The antenna
array and feeding line are placed on the same dielectric substrate. The parallel
reflector is at distance λ0/4 from antenna array.
MTI1.2
ANTENNA
STRUCTURE ON CYLINDRICAL OBJECT WITH A QUASI-ISOTROPIC RADIATION PATTERN
Aleksandar
Nešić, IMTEL
Komunikacije, Belgrade,
Serbia
Dušan Nešić, ICTM-CMT,
University of Belgrade, Belgrade, Serbia
Introduced .innovation is an antenna system on a
cylindrical object with a spherical (isotropic) radiation pattern. In that
way, radio communication and detection from all directions between
cylindrical object and corresponding stations are obtained without any
switching. The highest deep deviation from the spherical shape (0 dBi) is
around 4 dB.
MTI1.3
WIDEBAND PRINTED ANTENNA FOR K BAND INTEGRATED WITH A LOW
NOISE AMPLIFIER
Siniša
Jovanović, IMTEL Komunikacije, Belgrade, Serbia
Zoran
Živanović, IMTEL Komunikacije, Belgrade, Serbia
This paper features the design, practical realization and
measured results of an active receiving antenna developed for field testing
of microwave links operating at K frequency range. The proposed solution
combines a printed antenna array with a low noise amplifier to obtain a
convenient small size, low power‑consuming device with an overall gain
of more than 43 dBi. Since the required frequency range was more than 10%
wide a pentagonal dipole was used instead of a standard patch antenna as the
radiating element of a printed antenna array. Also, to enable integration
with a low noise amplifier the pentagonal antenna array is for the first time
realized on a hydrocarbon ceramic substrate instead of a previously used teflon-fiberglass
substrate. Excellent matching in a very wide frequency range is achieved by
employing wideband Klopfenstein-type impedance transformers as major building
parts of the antenna array feeding network. Finally, a thick antenna cover in
addition to robust protective case is used to ensure the sturdiness of the
device for everyday usage for field testing. The final measurements of the
overall antenna characteristics showed very good results that are in very
good accordance with the theoretical predictions.
MTI1.4
EXPERIMENTAL
VERIFICATION OF THE INTEGRAL CYLINDRICAL TLM MODEL OF A PROBE-FED MICROSTRIP
CIRCULAR ANTENNA
Tijana
Dimitrijević, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Jugoslav
Joković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Nebojša
Dončov, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Bratislav
Milovanović, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Kurt Blau,
Technical University of Ilmenau, Ilmenau, Germany
In this paper,
an accuracy of the integral cylindrical TLM method has been experimentally
verified on an example of a coax-fed circular patch antenna fabricated on the FR4 substrate.
Obtained results are also compared with corresponding results reached by the
rectangular grid based TLM method as well as with analytical results based on
the cavity model.
MTI1.5
UNMITERED BEND
STRUCTURE – MODELING AND ANALYSIS BASED ON A COMBINED WAVE DIGITAL/ FULL-WAVE
ELECTROMAGNETIC APPROACH
Biljana
Stošić, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Nebojša
Dončov, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
An efficient technique for modeling of microstrip unmitered
bend structure is reported in this paper. The described technique uses a full
wave electromagnetic approach to generate a transmission line model of bend
discontinuity from its S-
parameters. The model is then incorporated into an equivalent wave digital
network representing the complete microstrip bend structure. Two single
right-angled microstrip bend structures, with 900 and 450
bend discontinuities are considered here in order to validate the presented
combined wave digital/EM approach technique.
Session MTI2
Chairs: Vera
Marković, University of Niš,
Serbia
Olivera
Pronić-Rančić,
University of Niš, Serbia
Wednesday, June 4, Room 4, 8:30
MTI2.1
ANN APPROACH FOR THE ANALYSIS OF THE RESONANT FREQUENCY
BEHAVIOR OF RF MEMS CAPACITIVE SWITCHES
Zlatica
Marinković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Tomislav
Ćirić, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Olivera
Pronić-Rančić, Faculty of Electronic Engineering, University
of Niš, Niš, Serbia
Taeyoung Kim, TU München, Lehrstuhl für Hochfrequenztechnik, Arcisstr. München, Germany
Marija Milijić, Faculty of Electronic Engineering,
University of Niš, Niš, Serbia
Vera
Marković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Larissa Vietzorreck, TU München, Lehrstuhl für Hochfrequenztechnik,
Arcisstr. München, Germany
RF MEMS switches are widely applied in modern communication
systems. For the design of circuits and systems containing these switches it
is necessary to have their accurate and reliable models. In this paper it is
illustrated how RF MEMS switch models based on artificial neural networks can
be used for the analysis of the RF MEMS capacitive switch resonant frequency
behavior. The resonance frequency is controlled by changing the switch bridge
dimensions, to have a scalable model relating geometry and switch performance
parameters turns out to be very useful during the design of a system or
circuit containing RF MEMS switches.
The change of resonant frequency with the tolerances in the bridge
dimension fabrication is studied as well. The proposed approach allows the
designers to analyze the switch behavior in a significantly shorter time
compared to the same analysis performed with standard electromagnetic
full-wave simulators.
MTI2.2
EXTRACTION OF INTRINSIC
NOISE PARAMETERS OF MICROWAVE FETS BASED ON ANN
Vladica
Đorđević, Innovation centre of advanced technology, Niš, Serbia
Zlatica
Marinković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Olivera
Pronić-Rančić, Faculty of Electronic Engineering, University
of Niš, Niš, Serbia
Vera
Marković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
A procedure of extraction of intrinsic noise parameters of
microwave FETs based on artificial neural networks is presented in this paper.
The reason why the extraction of
these parameters is carried out, is to make simplified the entire procedure
for determining the noise model parameters. A neural network is
trained to predict noise parameters of intrinsic circuit for given equivalent
circuit parameters, transistor total noise parameters, frequency and ambient
temperature. The detailed validation of the trained artificial neural network
was done by comparison of the transistor total noise parameters obtained
using extracted noise parameters of intrinsic circuit and measured transistor
total noise parameters.
MTI2.3
BEHAVIORAL
MODELING OF LOW NOISE AMPLIFIER FOR LTE SYSTEMS BASED ON RECURRENT NEURAL
NETWORKS
Jelena Mišić,
Faculty of Electronic Engineering, University of Niš, Niš, Serbia
Vera
Marković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Zlatica
Marinković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Djuradj Budimir, the
Wireless Communications Research Group,University of Westminster, London, UK
In this paper, the behavior model of a low-noise amplifier
for LTE receiver, based on a recurrent artificial neural network, is presented. The development of the
model has been performed in MATLAB software environment. For the modeling
purpose, the measured data of the device Мini Circuit ZFL-500 has been employed.
The input signal is an LTE signal with 3MHz channel bandwidth and carrier
frequency of 1960MHz. The output
characteristics obtained by the proposed model agree very well with measured
ones.
MTI2.4
ANALYSIS OF LINEARIZATION CIRCUIT IMPACT ON BROADBAND DOHERTY
AMPLIFIER PERFORMANCES
Aleksandra
Đorić, Innovation centre of advanced technology, Niš, Serbia
Nataša
Maleš-Ilić, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
Aleksandar
Atanasković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
The impact of linearization circuit on broadband two-way
Doherty amplifier for application in the frequency range 0.95-1.05 GHz is
considered in this paper. The carrier and peaking amplifiers in Doherty
configuration comprise Freescale’s transistor MRF281S LDMOSFET characterized
by the maximum output power 4W and the broadband combined matching circuits
with lumped elements and transmission lines. The linearization of the
amplifier is carried out by the second- and fourth-order nonlinear signals
that are extracted at the output of the peaking cell, adjusted in amplitude
and phase and fed at the input and output of the carrier cell in Doherty
amplifier. Since the extraction and injection of the signals for
linearization are carried out through the band-pass filters, the influence of
different types of filters on broadband Doherty amplifier performances is
analyzed. Additionally, the effects of linearization are considered for WCDMA
digitally modulated signal.
NUCLEAR ENGINEERING – NT
NTI1 Use of Nuclear Radiation
Chair: Selena Grujić, Faculty of Technical
Sciences, Novi Sad, Serbia
Tuesday, June 3, Room 2, 14:00
NTI1.1
QUASI-MINIMAL RESIDUAL METHOD FOR X-RAY SPECTRUM UNFOLDING
Jelena Stanković, Institute for Nuclear Sciences
Vinča, Lab for protection against radiation and environmental
protection, P.O. Box 522 Vinča, 11001 Belgrade, Serbia
Predrag Marinković, School of Electrical Engineering,
University of Belgrade, Bulevar kralja Aleksandra 73, 11020 Belgrade, Serbia
Spectrum unfolding is crucial to characterize primary
X-ray spectrum i.e. spectrum before any interactions in spectrometer
material. This paper is presenting results obtained when Quasi-minimal
Residual Method (QMR) is used for X-ray spectrum unfolding. Pulse
amplitude spectrum was first acquired with CdTe spectrometer and after that
unfolded. Energy response of CdTe spectrometer was simulated using Monte Carlo
program MCNP5. Unfolded spectrum agreed well with reference spectra
calculated via semi-empirical method, still results showed that correction
due to carrier trapping must be incorporated in Monte Carlo calculation.
NTI1.2
OCCUPATIONAL DOSES IN INTERVENTIONAL CARDIOLOGY
Jelica Kaljević, Vinca Institute of Nuclear Science,
University of Belgrade, P.Box 522 Vinča, 11001Belgrade, Serbia
Olivera Ciraj-Bjelac, Vinca Institute of Nuclear Science,
University of Belgrade, P.Box 522 Vinča, 11001Belgrade, Serbia
Jelena Stanković, School of Electrical Engineering,
University of Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Danijela Aranđić, School of Electrical Engineering,
University of Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Predrag Božović, School of Electrical Engineering,
University of Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Vojislav Antić, School of Electrical Engineering,
University of Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Occupational doses from interventional procedures are the
highest among medical staff using x-rays. The aim of the study was to
evaluate the level of the occupational doses in modern international
cardiology procedures in Serbia. Effective dose levels were analyzed by means
of two methods. One method consisted of measuring ambient doses in terms of
H*(10), and the another one of measuring personal dose equivalent,
Hp(10). For measuring H*(10) ionization chamber was used. Thermoluminescent
dosimeters (TLD) were used for Hp(10) assessment. Method for measuring H*(10)
gave higher results than TLD method, but both results showed values that are
in accordance with annual dose limits for the occupational exposure.
The efficiency of personal protective tools was also assessed. The presented
results showed that one third of monitored persons are not wearing TLDs
correctly and that additional training of the medical staff should be
conducted.
ROBOTICS AND FLEXIBLE AUTOMATION – RO
ROI1 Design, kinematics, dynamics
Chair: Huosheng Hu, University of Essex, UK
Wednesday, June 4, Room 3, 8:30
ROI1.1
DEVELOPMENT OF EXPERIMENTAL PLATFORM FOR RESEARCH IN ROBOTS
HAVING COMPLIANT JOINTS
Veljko Potkonjak, School of Electrical Engineering,
University of Belgrade, Bulevar Kralja Aleksandra 73, 11020 Belgrade, Serbia
Branko Lukić, School of Electrical Engineering,
University of Belgrade, Bulevar Kralja Aleksandra 73, 11020 Belgrade, Serbia
Zaviša Gordić, School of Electrical Engineering, University
of Belgrade, Bulevar Kralja Aleksandra 73, 11020 Belgrade, Serbia
Predrag Milosavljević, Automatic Control Laboratory,
Swiss Federal Institute of Technology Lausanne (EPFL), Lausanne,
Switzerland
This paper describes the development and testing of the
experimental platform for the control of robots having compliant joints.
Testing consists of the three smaller tasks. The tasks were designed to test
whether the realized platform meets all the necessary requirements to
properly control the robot. Each task represents a segment required for the
implementation of robot control software (communication with the hardware,
the exchange of data between the HOST PC and the controller, the processing
of data in real time). In particular, attention has been paid to the rational
use of resources on the controller. We used a FPGA Spartan 3 of the NI cRIO
9074 chassis. We controlled position of 5 DC drives. Reference is given by
turning the potentiometer, entered from the keyboard, or loaded from a text
file. The problems for each method of assigning references and adopted
solutions are presented.
ROI1.2
COMPLIANT BEHAVIOR OF REDUNDANT ROBOT ARM – EXPERIMENTS WITH
NULL-SPACE COMPLIANCE
Nikola Lukić, Faculty of Mechanical Enginering, Belgrade
University, Kraljice Marije 16, 11120 Belgrade, Serbia
Ivan Danilov, Faculty of Mechanical Enginering, Belgrade
University, Kraljice Marije 16, 11120 Belgrade, Serbia
Petar B. Petrović, Faculty of Mechanical Enginering,
Belgrade University, Kraljice Marije 16, 11120 Belgrade, Serbia
In this paper we present recent research activities related
to impedance control of redundant robot arms. Redundancy is widely studied
for decades theoretically and very seldom, achieved theoretical results are
proved by experiments. Therefore, deep understanding the principles of
accurate motor coordination with redundant degrees of freedom still remains a
challenging problem. This is particularly true for new research in highly
complex robotic systems such as humanoid robots, and especially a new generation
of industrial robots, i.e., industrial humanoids, which possesses humanoid
properties found in service robotics, but with one very important additional
requirement – high precision and accuracy. High precision and accuracy are
indispensable property for application of humanoid robots in industrial
environment, performing various tasks on manufacturing (mostly assembly)
lines. Redundancy resolution is typically based on use of the Jacobian
pseudoinverse techniques, with local null-space optimization in order to
determine the inverse kinematics transformation. In a number of research
studies, kinematic redundancy is resolved at the velocity, acceleration or
torque levels in order to achieve desired TCP velocity, acceleration, and
force profiles. Achieved theoretical results are proved typically by
simulation studies, using simple TCP movements, limited to geometrical
primitives only. Moreover, stability of derived algorithms is incompletely
studied as well as influence of inevitable modeling errors of the robot
dynamics. For that reason we have developed a dedicated platform for
experimentation based on 7 d.o.f. SIA 10F Yaskawa robot arm and appropriate
sensory system for displacement and force measurement and acquisition. The
robot controller YASKAWA FS100 is of newest generation, with very fast servo
loops, down to 1 ms for all of controlled degrees of freedom. In addition, a
special development environment for Yaskawa FS100 controller, named MotoPlus,
is used to create application programs which can be executed as a tasks for
the main CPU of the robot controller, with ability to access very basic
control functions and variables / data in real-time. API (Application Program
Interface) routines which are included in the MotoPlus environment allows access
to fast data transmission (via Ethernet or RS232C serial) and control of all
robot functionalities. Such open control system enables extensive
experimentation with fully controllable redundant robot arm and evaluation of
various theoretically developed redundancy resolution schemes / theories,
including those one which are focused to the generalized / extended impedance
control and redundancy resolution algorithms (modifying / control of the
null-space is incorporated) which is robust against modeling errors and
unforeseen disturbances arising from contact forces. This paper is
particularly focused on our ongoing research activities related to the
compliant motion / behavior issues in the robot null-space and some of the
aspects related to overall performances of the redundant robot arm when the
dynamics of task space and null motions are coupled together.
ROI1.3
DUAL-ARM ROBOT TRACKING OF THE MOVING TARGET USING THE
ALGORITHM OF INVERSE KINEMATICS
Marija Tomić, Robotics Laboratory, Mihailo Pupin Institute,
University of Belgrade, Volgina 15, 11000 Belgrade, Serbia and
School of Electrical Engineering in Belgrade, Buleva kralja Aleksandra
73, 11000 Belgrade, Serbia
Duško Katić, Robotics Laboratory, Mihailo Pupin Institute, University of
Belgrade, Volgina 15, 11000 Belgrade, Serbia
The tendency of scientists to integrate robots into human
environments has led to a boom in the development of more sophisticated
robotic system inspired by the human skills like dual-arm robotics
manipulation. In this paper a conventional algorithm of inverse kinematics
(IK algorithm) is proposed for natural imitation of human upper body motion.
Using the model of dual arm robot and IK algorithm inverse kinematics, the
two tasks are analyzed: following the moving target by predefined trajectory
and tracking stochastic moving points. For designing optimization algorithms of monitoring, it is necessary to
add some more criteria in inverse kinematics algorithm. Proportional
navigation algorithm is chosen as possible solution for monitoring the moving
point, which has shown good tracking performance of moving point.
ROI1.4
D-DECOMPOSITION METHOD FOR STABILIZATION OF INVERTED PENDULUM
USING FRACTIONAL ORDER PD CONTROLLER
Petar D. Mandić,
Faculty of Mechanical Engineering, University of Belgrade, 16 Kraljice
Marije, 11120 Belgrade, Serbia
Mihailo P. Lazarević, Faculty of Mechanical Engineering, University of Belgrade,
16 Kraljice Marije, 11120 Belgrade, Serbia
Tomislav B. Šekara, Faculty of Electrical Engineering,
University of Belgrade, 73 Bulevar kralja Aleksandra, 11120 Belgrade, Serbia
This paper deals with stability problem of inverted
pendulum controlled by a fractional order PD controller. D-decomposition
method for determining stability region in controller parameters space is
hereby presented. The D-decomposition problem for linear systems is extended
for linear fractional systems and for the case of linear parameters
dependence. Knowledge of stability regions enables tuning of the fractional
order PD controller.
ROI1.5
THE CONCEPT OF
RECONFIGURABLE ROBOTIC PLATFORM
Ilija
Stevanović, Robotics Laboratory,
Mihailo Pupin Institute, University of Belgrade, Volgina 15, 11000
Belgrade, Serbia
Improvement in
robotics and in movement of robot platforms is a continuing pursuit. Over the
past years there is a noticeable progress in development of existing mobile
robot platforms designed to move over uneven terrains. Reason for this trend is increased demand
for this systems is application in different critical situations such as:
(search and rescue of earthquake
survivals, firefight situation
assessment and search and rescue injured people, situation assessment and decontamination of nuclear power plants
after incidents and in different environmental accidents where direct human
access is too dangerous or
impossible). The robotic platform should be able to move between areas of
interest quickly and safely. Wheeled
transportation can be characterized by
greater speed and efficiency,
while (biologically inspired) articulated leg
transportation can be
characterized by greater flexibility
for movement over
complex terrain. Wheeled robot
platforms can move quickly, but can have difficulty on uneven terrain,
while articulated legs or other different
types of motion can negotiate
the uneven terrain, but
can have difficulty
with speed. Such
wheeled transport can have
limited mobility and
behavior due to complex
environment and lack
of adaptability to unpredictable terrain. The solution is a
new type of vehicles which inherits both advantages of legged and wheeled
vehicles, namely the high adaptive
capabilities of legs and the high velocity and payload of the wheels. In
order to deal with
the rough terrains
of planetary surfaces, researchers put
most of the
efforts in designing
new structure of rover body, but give less attention to new types
of reconfigurable mechanism
for wheels and
or easy changeable motion systems that depends of type of terrain that
the robotics platform need to move across.
ROI2. Bio-Inspired Systems & Humanoid
Robotics
Chair: Stevo Bozinovski, South Carolian State
University, USA
Thursday, June 5, Room 3, 8:30
ROI2.1
SVM REGRESSION-BASED
COMPUTED TORQUE CONTROL OF HUMANOID ROBOT REACHING TASK
Srđan
Savić, Faculty of Technical Sciences, University of Novi Sad, 6 Trg
Dositeja Obradovića, 21000 Novi Sad, Serbia
Mirko
Raković, Faculty of Technical Sciences, University of Novi Sad, 6 Trg
Dositeja Obradovića, 21000 Novi Sad, Serbia
Milutin
Nikolić, Faculty of Technical Sciences, University of Novi Sad, 6 Trg
Dositeja Obradovića, 21000 Novi Sad, Serbia
Branislav Borovac,
Faculty of Technical Sciences, University of Novi Sad, 6 Trg Dositeja
Obradovića, 21000 Novi Sad, Serbia
This paper
presents the application of Support Vec- tor Machine regression (SVR) for
realization of the robot arm reaching task. Robot arm with 4 DOFs (Degrees Of
Freedom) was used in simulation for solving trajectory tracking problem. Only
position of the target point was defined, not the orientation, so the system
had one redundant DOF for the considered task. Inverse kinematics was solved
at the velocity level and redun- dancy was exploited to keep the joints away
from their mechani- cal limits. The advantage of the proposed approach over
the well- established control techniques like computed torque control and
inverse dynamics control lies in the fact that with SVR it is not necessary
to know exact dynamic model which is actually never realizable in practice,
due to complex friction models, actuator dynamics and unmodeled
nonlinearities. In this paper ε-SVR was used to estimate nonlinear
dynamic model of the robot arm. Outputs from the SVR are learned driving
torques which are used as nonlinear feedforward control in conjunction with
PD feedback regulator to achieve positioning task in presence of external
disturbance.
ROI2.2
3D BIPED GAIT
REALIZATION USING INVERTED PENDULUM ANALOGY
Miloš D.
Jovanović, Mihailo Pupin Institute, University of Belgrade, Volgina 15,
Belgrade Serbia
Veljko Potkonjak, Faculty of Electrical engineering,
University of Belgrade, Bulevar
kralja Aleksandra 73, Belgrade, Serbia
One common
interpretation of bipedal walking is equivalence with inverse pendulum. In a
number of attempts, the biped is represented by a simple system with only one
to complex with the pendulum of the pendulum, in the general case by one
extremity by a pendulum. The solid is suitable for describing the pendulum
course, are not suitable for other modes describe bipedal motion such as, for
example, scoring. For purposes of describing the runs are used which in
addition to the mass of the pendulum and a spring, which provides a variable
longitudinal dimension. The simplest solution is the equivalence of a single
pendulum with a concentrated mass at the top, a constant length of the
pendulum and the connection at the point of support which has two degrees of
freedom. If for such a three-dimensional pendulum set of dynamic equations,
the globbing trajectory of the pendulum's center of mass, we can determine
the motion along the axis of the system set out in the rest point.
ROI2.3
ONE OF THE SOLUTIONS FOR CONVERSION HUMAN TO HUMANOID MOTION
Marija Tomić, Mihailo Pupin Institute, University of
Belgrade, Volgina 15, Belgrade, Serbia
Christine Chevallereau, Institut de Recherche en
Communications et Cybernetique de Nantes, Nantes, France
In the recent years, the focus of research in the field of
humanoid robotics is an imitation of human movement. The development of
technology has led to various systems for the capture of human motion. One of
the systems for recording 3D human motion and marker-based motion capture
system. This paper presents an algorithm of conversion the recorded human
movements to the robot’s movements. The basic ideas for the human to humanoid
motion conversion are to scale the robot segment dimensions to the
human limbs size, attach the model of robot Virtual Markers such that during
the movements their position and orientation evolution match the Real Markers
evolution attached to the actor and following the position of actor’s
joints obtained from capture motion systems. Based on these principles the
optimization algorithm gives the humanoid imitation of motion in joint space.
As a model of humanoid the ROMEO robot is used. The algorithm was tested on
measurements from several experiments and comparative analysis of the results
was performed.
ROI2.4
MUSCLE MODELS FOR ACCURATE
SIMULATION OF HUMAN MOVEMENTS
Kosta
Jovanović, Faculty of Electrical
engineering, University of Belgrade, Bulevar kralja Aleksandra 73, Belgrade, Serbia
Jovana
Vranić, Faculty of Electrical engineering, University of Belgrade, Bulevar kralja Aleksandra 73, Belgrade,
Serbia
Safe human-robot
interaction is one of the key issues in future service robotics. Numerous
researchers have already invested a lot of effort to design passively
compliant robots and to investigate active compliance through control
algorithms. On the top of that, human-like robots are expected to move, act
and fit into fully human adapted environment. To exploit human movement
patterns and design human like actuators we firstly need to understand how
humans move. This paper presents different muscle models (Hill’s and Huxley’s
model), elaborates their features and demonstrates
the trade-off between their accuracy and efficiency of computer
simulations. Comparative analysis of the models is done, and comparison of
simulation results to experimentally measured human movement is carried
out.
ROI2.5
LIGHT WEIGHT ARMS
- AN OVERVIEW
Svemir Popić,
Institute Mihailo Pupin, Robotics laboratory, Volgina 15, 11060
Belgrade, Serbia
Industrial
robots - manipulators are indispensable devices in carrying out the tasks of
accurate and repeatable of his end effectors positioning. For this reason it
is extremely important to have a system with high rigidity and power
capacity, able to quickly and reliably manipulate with his body weight and
payloads. On the other hand, the Light weight arms have a new complex task.
The own thought mass in relation to the payload must be approximately equal,
with limitation in installed power and with sufficient system dynamic. This
is achieved by a new constructive solutions as well as implementation of new
algorithms and control systems.
ROI2.6
TOWARDS BUILDING OF LIGHTWEIGHT ROBOT ARM OF ANTHROPOMORPHIC
CHARACTERISTICS
Aleksandar
Rodić, Institute Mihailo Pupin, Robotics laboratory, Volgina 15, 11060
Belgrade, Serbia
Branko
Miloradović, Institute Mihailo Pupin, Robotics laboratory, Volgina 15,
11060 Belgrade, Serbia
Đorđe
Urukalo, Institute Mihailo Pupin, Robotics laboratory, Volgina 15, 11060
Belgrade, Serbia
The subject of
research in this paper does not concern with development of new lightweight
robot arm of better technical characteristics than the existing industrial
robots available at the market. The aim is setting of the new design concept
of a bionic robot arm, based on the priciples known from human bio-mechanics,
that leads towards achieving better antropomorphic characteristics of the
robot. The novelties to be presented in this paper regard to decresing of the
overall robot weight, increasing the payload fraction (the mass and payload
ratio) and introducing of the spherical joint in the shoulder instead of
commonly used cylindric joints in sequence.
Also, the new proposed robot-arm structure represnets a redundant,
over-actuated robot mechanism such as human arm with muscles that are more
numerous than the number of degrees of freedoom. We expect that one such
approach will lead to decresing of the robot mass as well as saving energy
making system more reliable and efficient. In parallel, we will take care to
keep the required high precision and reapitibility of the robot mechanism as
well as to enable robot to be enough speed in order to be implemented with
service robots (e.g. dual arms robotic system).
ROI3. Control, Technology and Applications
Chair: Aleksandar Rodić, Mihailo Pupin
Institute, Belgrade, Serbia
Thursday, June 5, Room 3, 10:30
ROI3.1
SURVEY OF VIRTUAL LABORATORIES AND VIRTUAL ENVIRONMENTS WITH EMPHASIS ON APPLICATIONS IN
ROBOTICS
Veljko Potkonjak, School of Electrical Engineering, University
of Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Vladimir M. Petrović, School of Electrical Engineering,
University of Belgrade, 73 Bulevar kralja Aleksandra, 11020 Belgrade, Serbia
Due to the high expenses that modern laboratories and
learning facilities require, in recent years much attention has been paid to
various forms of virtual education. This paper deals with virtual
laboratories and virtual environments in education. We will first give an
overview of existing virtual laboratories in STEM (Science, Technology,
Engineering, Mathematics) fields. Since our focus is on robotics, special
attention is given to virtual laboratories developed in this field. Also, an
analysis of the exisiting VE (virtual environment) technologies is provided.
Finally, Virtual Laboratory for Robotics, developed at ETF is fully
described, illustrating our viewpoint on the entire idea of virtual
laboratories.
ROI3.2
THE COMPARISON BETWEEN THE REAL AND THE SCALED MODEL OF THE
CPR SYSTEM
Ljubinko Kevac, School of Electrical Engineering, The
University of Belgrade, Bulevar Kralja Aleksandra 73, 11000 Belgrade,
Serbia; Innovation center, School of
Electrical Engineering, The University of Belgrade, Bulevar Kralja Aleksandra
73, 11000 Belgrade, Serbia
Mirjana Filipović, Mihajlo Pupin Institute, University
of Belgrade, Volgina 15, 11060 Belgrade, Serbia
Ana Djuric, Wayne State University , 4855 Fourth St. Detroit,
MI 48202, USA
In this paper, we present one configuration of the
Cable-suspended parallel system, ie. CPR system, for the workspace
surveillance. We have designed two versions of this system and we have
compared these systems. First, we have designed a real version of this
system, ie. “the big system”, which presents a starting idea. Then, we have
designed a smaller (scaled) replica of this system, ie. “the small system”.
Afterwards, we have tested the both systems for similar conditions and for
adequate motors; and then we have compared results acquired from both
systems. From these results we have made adequate concluding remarks and
plans for future development.
ROI3.3
COMPLEXITY OF THE ELASTIC S-TYPE CABLE-SUSPENDED PARALLEL
ROBOT
Mirjana Filipović, Mihajlo Pupin Institute, University
of Belgrade, Volgina 15, 11060 Belgrade, Serbia
Ana Djuric, Wayne State University , 4855 Fourth St. Detroit,
MI 48202, USA
Ljubinko Kevac, School of Electrical Engineering, The
University of Belgrade, Bulevar Kralja Aleksandra 73, 11000 Belgrade,
Serbia; Innovation center, School of
Electrical Engineering, The University of Belgrade, Bulevar Kralja Aleksandra
73, 11000 Belgrade, Serbia
The presence of the elastic property in the ropes
significantly increases the complexity of the eSCPR system (elastic S-type
Cable-suspended Parallel Robot). The significance of the eSCPR system
modeling is the relation between the motor angular position and the elastic
deformation of the corresponding rope which is defined by the fictitious
coordinates. This novel procedure is named ED+M method, which means Elastic
Deformations plus Motor motion. The Jacobian matrix of the eSCPR system
relates the velocities of the external coordinates with the velocities of the
fictitious coordinates. The relation between the fictitious elastic load
moment and the external force, which is calculated using the Lagrange
principle of virtual work, is expressed with the Jacobian matrix. The motors
loads are expressed from the same calculation. Three different case studies
of the eSCPR were analyzed and presented.
ROI3.4
MODELING AND CONTROL OF CAR HANDLING BOX SYSTEM
Zaviša Gordić,
PhD student, School of Electrical Engineering, University of Belgrade,
Bulevar Kralja Aleksandra 73, 11020 Beograd, Serbia
Kosta Jovanović, PhD student, School of Electrical Engineering, University
of Belgrade, Bulevar Kralja Aleksandra 73, 11020 Beograd, Serbia
This paper presents two approaches to Car Handling Box
system modeling and design of a controller.
With a goal of obtaining adequate model of a process, Newton and
Lagrange based methods were used separately. Both models were tested under
same conditions and more comprehensive model was chosen for further work.
Controller was designed in a form of cascade consisting of two real PD
regulators in conjunction with low-pass filters. Performances of closed loop
system were tested for robustness to disturbances, measurement noise and
parameter uncertainty.
COMPUTERS – RT
RTI1. Computers
Chair: Claudio Moraga, Ivan Milentijević
Tuesday, June 3, Room 1, 15:00
RTI.1.1
PERFORMANCE COMPARISON OF RAID-1, RAID-0 AND SINGLE DISK ON
OPERATING SYSTEM MS WINDOWS 7
Borislav
Djordjević, Institute Mihajlo Pupin, University of Belgrade,
Serbia
Valentina
Timčenko, Institute Mihajlo Pupin, University of Belgrade, Serbia
Nikola
Davidović, Faculty of Electrical Engineering at University of
East Sarajevo, East Sarajevo, RS BiH
This paper presents results of performance comparison for
two different RAID configuration disks, RAID-0 and RAID-1, using CrystalMark
software and HD Tune Pro tool. Special accent is on the analysis of results
and on comparison to performances of a single disk. Result analysis is based
on mathematical model of write and read operations for chosen RAID
configuration and for every single disk. In order to gain validity and comparability
of the results, all tests are conducted under the same conditions, i.e. in
same environment and partition.
RTI.1.2
EXPLORATORY SPATIO-TEMPORAL ANALYSIS TOOL FOR LINKED DATA
Dejan
Paunović, Institute Mihajlo Pupin, University of Belgrade, Serbia
Valentina Janev, Institute
Mihajlo Pupin, University of Belgrade, Serbia
Vuk Mijović, Institute
Mihajlo Pupin, University of Belgrade, Serbia
Linked Data provides a publishing paradigm in which not
only documents, but also data, can be a first class citizen of the Web,
thereby enabling the extension of the Web with a global data space based on
open standards - the Web of Data. This paradigm has been used in an
increasing number of data stores in recent years, including data stores with
spatio-temporal data. Analysis of spatio-temporal data is not a
straightforward task due to the complexity of the data structures, together
with the representation and manipulation of the data involved. This paper
considers challenges and directions for modelling spatio-temporal data and
describes the first prototype of the Exploratory Spatio-Temporal Analysis
tool for Linked Data.
RTI.1.3
CONCERN-ORIENTED API DESIGN FOR IMPROVED USABILITY AND
MAINTAINABILITY
Žarko Mijailović, Faculty of Electrical Engineering,
University of Belgrade, Serbia
Dragan Milićev, Faculty of Electrical Engineering,
University of Belgrade, Serbia
Inheritance hierarchies of mainstream application
programming interfaces (API) are too often developed in contrast with
usability experience of their users. Very often, inheritance hierarchies are
too deep, too complex, and hard to explore. Although many design guidelines
exist, popular APIs provide diverse design configurations of their
inheritance trees. Moreover, it seems that there is no agreement in terms of
how exactly to use interfaces in API design. We propose a novel approach to
API design with improved API usability and maintainability of the client
code. The goal is to adjust API design to fit programmers’ needs. More
precisely, our approach is oriented towards high-level concerns of the API
users. It adopts benefits of interface-based programming style, while at the
same time avoids its common pitfalls.
RTI.1.4
EFFICIENT COMPUTATION OF THE OPTIMAL REED-MULLER SPECTRUM ON
GRAPHICS PROCESSORS
Miloš Radmanović, Faculty of Electronic Engineering,
University of Niš, Niš, Serbia
Dušan B. Gajić, Faculty of Electronic Engineering,
University of Niš, Niš, Serbia
Radomir S. Stanković, Faculty of Electronic Engineering, University of Niš, Niš,
Serbia
The Reed-Muller transform is a useful mathematical tool for
the analysis, synthesis, and optimization of Boolean functions. The number of
spectral coefficients can be reduced by selecting either positive or negative
literals for variables, which leads to the fixed-polarity Reed-Muller (FPRM)
spectra. For practical applications, it is often useful to be able to
efficiently compute these spectra in order to select the spectrum with the
minimum number of non-zero coefficients (the optimal spectrum). This paper proposes
techniques for the efficient computation of optimal FPRM Reed-Muller spectrum
for the given function, by using graphics processing units (GPUs). We present
the mappings of computation algorithms based on exhaustive search and dual
polarity route to the GPU. The algorithm is implemented both as sequential
C++ code and using the CUDA (Compute Unified Device Architecture) framework
for the development of parallel GPU applications. These approaches are
experimentally verified and compared. It is found that the computations of
the optimal spectrum on a GPU are quite efficient in terms of the computation
time.
TELECOMMUNICATIONS – TE
ТЕI1 Telecommunications
Chair: Igor Radušinović, Faculty of
Electrical Engineering in Podgorica, Montenegro
Thursday, June 5, Room 1, 8:30
TEI1.1
UPDATING OF PARALLELIZED IPv6 LOOKUP ALGORITHMS
Nataša Maksić, Innovation Center of the School of
Electrical Engineering, Belgrade, Serbia
Zoran Čiča, School of Electrical Engineering,
Belgrade, Serbia
Aleksandra Smiljanić, School of Electrical Engineering,
Belgrade, Serbia
The transition from IPv4 to IPv6 is inevitable since the
IPv4 address space is becoming depleted, while the number of Internet users
is growing. Memory requirements of IP lookup algorithms increase with the
address length, and, thus limit the scalability and speed of these
algorithms. We proposed earlier the BPFL algorithm which has reduced memory
requirements. In this paper, we examine updating of BPFL IPv6 lookup tables,
and compare it with the benchmark lookup algorithm, POLP, which is based on a
conventional trie lookup algorithm sped up using parallelization and
pipelining. First, we derive expressions of the worst-case memory
requirements for POLP and BPFL, and confirm the superiority of BPFL in terms
of scalability. Then, we examine updating complexities of these two lookup
algorithms. It will be shown that updating is faster in POLP, and, therefore,
less intrusive. For large lookup tables, updating speed in BPFL remains
acceptable.
TEI1.2
INTEGRATION OF QUAGGA ROUTING SOFTWARE AND NetFPGA-10G PLATFORMS
Saša Takov, Inovation Center, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
Jelena Seović, Inovation Center, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
Jelena Veličković, Inovation Center, School of
Electrical Engineering, University of Belgrade, Belgrade, Serbia
Aleksandra Smiljanić, School of Electrical Engineering,
University of Belgrade, Belgrade, Serbia
Modern Internet services require link and routers providing
high bandwidths. In order to help promote education and competition in the
area of high-capacity routers, several routing platforms have been developed, most prominent ones
being the NetFPGA-10G platform and the Quagga open-source routing software.
In this paper we will describe integration of these two platforms, and its
validation. Our test will show that the IP lookup table calculated by the
central processor using the Quagga software can be successfully transported to
the QDR-II memory on the NetFPGA board. This integration is an important step
for the Internet router design since it has not been presented in the
literature before, while the code provided by the NetFPGA was closed and
inflexible. We used the proposed design to implement a high-speed routing environment that deploys a novel lookup algorithm
which was developed in our lab.
TEI1.3
EVALUATION OF THE OPEN-SOURCE IMPLEMENTATION OF PIM-SM
Jelena Veličković, Inovation Center, School of
Electrical Engineering, University of Belgrade, Belgrade, Serbia
Jelena Seović, Inovation Center, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
Aleksandra Smiljanić, School of Electrical Engineering,
University of Belgrade, Belgrade, Serbia
PIM-SM is the routing protocol which describes the exchange
of the network topology information in order to construct logical multicast
trees. Multicast traffic is routed along these multicast trees in order to
more efficiently utilize the network resources. In this paper, we evaluate
capabilities of the open-source implementations which allow routing of
multicast traffic: Quagga and Pimd.
TEI1.4
EVALUATION OF IS-IS IMPLEMENTED IN QUAGGA
Saša Takov, Inovation Center, School of Electrical
Engineering, University of Belgrade, Belgrade, Serbia
Aleksandra Smiljanić, School of Electrical Engineering,
University of Belgrade, Belgrade, Serbia
Routing protocols define rules according to which routers
exchange the network topology information. Based on this topology information
and a packet address, routers decide where to send incoming packets. IS-IS
routing protocol is the most scalable intradomain routing protocol. In this
paper, we will evaluate the IS-IS functionalities implemented in the Quagga
open-source software.
TEI1.5
ANALYSIS OF SOME ROUTING AND WAVELENGTH SELECTION TECHNIQUES
IN OPTICAL NETWORKS WITH DYNAMIC TRAFFIC DEMANDS
Suzana Miladić (Ph.D student), Faculty of Transport and
Traffic Engineering, University of East Sarajevo, Doboj, Bosnia and Herzegovina
Goran Marković, Faculty of Transport and Traffic
Engineering, University of Belgrade, Belgrade, Serbia
Fixed-alternate routing and some wavelength selection
techniques as two different parts of routing and wavelength assignment (RWA)
problem in wavelength-routed optical networks (WRON) with dynamic traffic
demands are focused in this paper. Such networks operate under the
wavelength-continuity constraint in which lightpaths are established for
connection requests between a source and a destination node pairs. For each
of these pairs several possible routes are determined in advance. We compared
first-fit and random-fit wavelength assignment algorithm for a given optical
network topology on the basis of blocking probability. The results are
analyzed by varying the values of network load per fiber link and the number
of alternate routes.
TEI1.6
POLICY BASED PERFORMANCE MANAGEMENT IN MOBILE AD HOC NETWORKS
Valentina Timčenko,
Institute Mihailo Pupin, University of Belgrade, Belgrade, Serbia
Slavica Boštjančič Rakas, Institute Mihailo
Pupin,University of Belgrade, Belgrade, Serbia
This paper addresses issue of automation in network
management for mobile ad hoc network (MANET). We have proposed a framework
for overall policy based management (PBM) solution applicable in MANET
environment. Proposed framework relies on the overlay management network
concept with two-layer architecture and it is independent of the underlying
network size, routing protocols, mobility models and mobile nodes speed.
Functional model of the proposed PBM solution consists of five entities
regarding management policies, quality of service, resources, security and
configuration
TEI1.7
LEVEL CROSSING RATE OF WIRELESS SYSTEM OVER GAMMA SHADOWED
Α-ΜMULTIPATH FADING CHANNEL
Časlav Stefanović, Faculty of Electrical
Engineering at University of Niš, Niš, Serbia
Dejan Milić, Faculty of Electrical Engineering at
University of Niš, Niš, Serbia
Danijel Đošić, Faculty of Electrical Engineering at
University of Niš, Niš, Serbia
Dragan Radenković, Faculty of Electrical Engineering at
University of Niš, Niš, Serbia
Petar Spalević, Faculty of Technical Sciences,
University of Priština, Kosovska Mitrovica, Serbia
Random α-μ
signal envelope with Gamma distributed power is considered. Received
signal experiences α-μ multipath
fading resulting signal envelope variation and Gamma shadowing resulting in
signal envelope power variation. Closed form expression for average level
crossing rate of signal envelope is derived. Obtained results can be used for
evaluation of closed form expression for average fade duration of wireless
communication system operating over Gamma shadowed α-μ multipath fading environment. Numerical results
are plotted graphically to show influence α-μ fading severity parameter and Gamma shadowing
several parameter on average level crossing rate.
TEI1.8
A SCALABLE LOAD BALANCING SOLUTION FOR A DIGITAL MULTIMEDIA
CONTENT DISTRIBUTION PLATFORM
Ognjen Joldžić, Faculty of Electrical Engineering Banja
Luka, Bosnia and Herzegovina
Zoran Djurić, Faculty of Electrical Engineering Banja
Luka, Bosnia and Herzegovina
The process of network convergence and the constant
advancement in the area of network performance have resulted in an increase
in availability of different types of digital multimedia content to network
subscribers. A modern, converged network presents a unified infrastructure
for content distribution, regardless of its type or other characteristics. In
this context, any implementation of digital video delivery networks has to
resolve several important issues in order to provide its users a satisfactory
level of service quality. One of the most important problems in this category
is the implementation of a responsive and scalable platform that would be
able to withstand high server loads caused by a large number of simultaneous
requests. This paper presents an overview of a load balancing solution which
is able to dynamically distribute the load across any number of nodes
configured within the distribution cluster. As a proof of concept, the
solution presented in this paper has been successfully implemented in a
Digital Multimedia Content Distribution System called GSTV.
ТЕI2 Telecommunications 2
Chair: Goran T. Đorđević,
Elektronski fakultet, Niš
Thursday, June 5, Room 1, 10:30
TEI2.1
FLEXIBLE FUTURE OF THE INTERNET: CHALLENGES AND RESEARCH
TRENDS
Invited Paper
K. R. Rao, Department of Electrical Engineering, University
of Texas of Arlington, USA
Zoran Bojković, University of Belgrade, Studentski trg
1, Serbia
Bojan Bakmaz, Faculty of Transport and Traffic Engineering,
University of Belgrade, Belgrade, Serbia
The Internet has been successfully deployed for several
decades due to its flexibility in operating using different physical media
and supporting various higher layer protocols and applications. However,
rapidly emerging applications with different requirements and implications
for future Internet design pose a significant set of challenges and research
problems. This work focuses on future Internet emphasizing current challenges
and research trends. We start with the comprehensive survey of future
Internet architecture, together with the information-centric principles.
Then, the new concept of video delivery is briefly presented. Finally,
self-management scenarios including adaptive resource management, energy
aware network management and cache management are described. The analysis
will enable the research community to focus on the key technologies to enable
the flexible future of the Internet.
TEI2.2
BER
PERFORMANCE ANALYSIS OF A MULTICHANNEL SATELLITE LINK EMPLOYING FDMA
Hana
Stefanović, High School of Electrical Engineering and Computer Science,
Vojvode Stepe 283, 11000 Belgrade, Serbia
Dejan
Milić, Faculty of Electronic Engineering, University of Nis, Aleksandra
Medvedeva 14, 18000 Nis, Serbia
Daniela
Milović, Faculty of Electronic Engineering, University of Nis,
Aleksandra Medvedeva 14, 18000 Nis
This
paper proposes some simulation models of a satellite based transponder
communication link, employing frequency division multiple access (FDMA). A
nonlinear power amplifier is assumed, while its impact on intermodulation (IM) products is
analyzed. Bit-Error-Rate (BER) performance in the downlink is given,
including the effect of Gaussian noise, IM distortion, intersymbol
interference (ISI) which is mainly due to the input and output multiplexing
filters, and adjacent channel interference (ACI) caused by increasing the
number of carriers and also varying the power amplifier back-off level.
ARTIFICIAL INTELLIGENCE – VI
VII1. Mathematical models of artificial intelligence
Predsedavajući: Aleksandar Perović
Thursday, June 5, Room 5, 8:30
VII1.1
FUZZY LOGIC AND FUZZY SET THEORY BASED EDGE DETECTION
ALGORITHM
Nebojša T. Perić, Group for Intelligent Systems, Faculty
of Mathematics, University of
Belgrade, Serbia
In this paper we will introduce a new way how to detect
edges in digital images. Edge detection is a fundamental part of many
algorithms, both in image processing and in video processing. Therefore it is
important that the algorithm is efficient and, if possible, fast to carry
out. The fuzzy set theory based approach on edge detection is good for use
when we need to make some kind of image segmentation, or when there is a need
for edge classification (primary, secondary...). One example that motivated
us is region labeling; this is a process by which the digital image is
divided in units and each unit is given a unique label ( sky, house, grass …
etc). To accomplish that, we need to
have an intelligent system that will precisely determine the edges of the
region. In this paper we will describe tools from image processing and fuzzy
logic that we use for edge detection as well as the proposed algorithm.
VII1.2
HEURISTIC APPROACHES FOR SOLVING THE PROBLEM OF
INFRASTRUCTURE EXPANSION OF A WIRELESS NETWORK
Miloš Šošić, Faculty of
Mathematics, University of Belgrade, Serbia
Zorica Stanimirović, Faculty of Mathematics, University of Belgrade, Serbia
Marko Šošić, Faculty of Mathematics, University of Belgrade, Serbia
In this paper, the problem of expansion of a wireless
network infrastructure is considered, which is of great importance in
telecommunications. The set of existing locations of base stations and
switching centers in a network is given, as well as the set of demand points
to be served. The problem is to expand the network by adding new locations
for base stations and switching centers, with respect to the location of the
demand points and given capacities of both base stations and switching
centers. The objective is to minimize the sum of cost of establishing new
base stations and switching centers and the costs of their inter-connections
and allocations to user nodes. Two heuristic methods - Iterated Local Search and
Ant Colony Optimization are proposed for solving the considered problem,
especially the cases of larger problem dimensions.
VII1.3
A NOTE ON WEIGHTED LOGICS
Obrad Kasum, Group for Intelligent Systems, Faculty of
Mathematics, University of Belgrade, Serbia
Aleksandar Perović, Group for Intelligent Systems,
Faculty of Mathematics, University of Belgrade, Serbia
Aleksandar Jovanović, Group for Intelligent Systems,
Faculty of Mathematics, University of Belgrade, Serbia
In this talk we will give a short survey of weighted logics
and their application in artificial intelligence. The emphasis will be on
propositional real-valued fuzzy logics and probability logics, the main
axiomatization issues, their relation with classical, modal logics and
theoretical computer science, as well as their application in various expert
systems.
VII1.4
SECURITY RISKS IN SYNCHRONIZATION ON-PREMISES ACTIVE
DIRECTORY WITH OFFICE 365 CLOUD PLATFORM
Dušan Stamenković, Singidunum University, Belgrade,
Serbia
Saša Adamović, Singidunum University, Belgrade, Serbia
Goran Šimić, Singidunum University, Belgrade, Serbia
Marko Šarac, Singidunum University, Belgrade, Serbia
Over the years, mail servers were moved a long way in accordance
with the development of new technologies. Microsoft’s started with basic mail
server integrated into the Windows Server operating system via the Exchange
Server until the Office 365 Cloud platform. Classic datacenter is comprised
almost inevitable with the local domain controller's and mail or Exchange
Server with a high level of security. In recent years, with the development
of Cloud technology and more affordable prices, there's a new resource type -
Hybrid Cloud. Hybrid Cloud is a combination of classical and modern (virtual)
datacenter, where the local datacenter kept on-premises active directory
while the Exchange server and Office365 is in Cloud on Azure platform far
away from your datacenter. Authors of
this paper reviewed safeguards and security mechanisms used in this
synchronization and data transfer as well, and the possible impact of
Heartbleed bug in implementation of TLS / SSL in Microsoft Azure platform.
VII1.5
AN OPEN-SOURCE SOLUTION FOR PROTECTING PHP SOURCE CODE
Nenad Ristić, Singidunum University, Belgrade, Serbia
Aleksandar Jevremović, Singidunum University, Belgrade,
Serbia
Mladen Veinović, Singidunum University, Belgrade, Serbia
Goran Shimic, Univerzitet Sinergija, Biljeljina, Bosnia and
Herzegovina
Protecting software code unwanted using, copying and
modifications is a pressing issue to many software developers. Current
mechanisms for protecting program source code are mostly working as
obfuscators, they are free, and are not providing any serious level of
protection. Solutions that are based on encrypting opcode are more secure,
nevertheless they are commercial and require closed-source proprietary PHP
interpreter's extension. Additionally, encrypted opcode is not compatible
with upcoming versions of interpreters which include re-buying encoders from
authors. Lastly, if extension source-code is compromised, all script encoded
by that solution are compromised too. In this paper we present a novel model
for free and open-source PHP script protection solution.
VII2 Learning techniques in artificial intelligence
Predsedavajući: Aleksandar Jevremović, Singidunum
University, Belgrade, Serbia
Thursday, June 5, Room 5, 10:30
VII2.1
ANALYZING BEHAVIOR OF WEB APPLICATIONS USERS AS A COMPONENT
OF AUTHENTICATION
Aleksandar Jevremović, Singidunum University, Belgrade,
Serbia
Marko Šarac, Singidunum University, Belgrade, Serbia
Milan Milosavljević, Singidunum University, Belgrade,
Serbia
Mladen Veinović, Singidunum University, Belgrade, Serbia
Authentication of Web applications users is generally based
on what user knows, and rarely by what user has, or what user is. Enabling
authentication based on what user has or is requires additional hardware
(smart card readers, fingerprint scanners, cameras, etc.) and therefore is
very difficult to implement. In this paper we are analyzing effectiveness of
using user's behavior as a component of authentication procedure. User
behavior is analyzed during the use of Web application for electronic
testing.
VII2.2
ON THE EVALUATION OF EMAIL SPAM FILTERS IN ADVERSARY
ENVIRONMENT
Lutfia Ramadan Milod, Singidunum University, Belgrade, Serbia
Milan Milosavljević, Singidunum University, Belgrade,
Serbia
This paper proposes a new approach to estimating the
robustness of e-mail spam filters in relation to possible attacker
strategies. The lower limit of filters vulnerability are obtained based on
the modification of learning
and/or test sets by the attacker, so
as to maximize performance drop measured by AUC criterion in the range of
high sensitivity values of the ROC
curve. Experimental results show that, email spam filters based on Random
Forest machine learning technique,
possesses superior robustness property, which is a novelty compared to the
results known so far.
VII2.3
APPLICATION OF HYBRID INCREMENTAL MACHINE LEARNING METHODS
TO ANOMALY BASED INTRUSION DETECTION
Vladislav Miškovic, Singidunum University, Belgrade, Serbia
Milan Milosavljević, Singidunum University, Belgrade,
Serbia
Saša Adamović, Singidunum University, Belgrade, Serbia
Aleksandar Jevremović, Singidunum University, Belgrade,
Serbia
Anomaly based intrusion detection systems can detect
computer systems misuse based on network and system behaviour where the type
of misuse isn’t previously known. Different machine learning methods explore
different hypothesis spaces, use different search strategies, different sets
of features, and are appropriate for different types of problems. Their
combination or integration usually gives better performance than using each
individual machine learning method on its own. Hybrid models can reduce
individual limitations of basic models and can exploit their different
generalization mechanisms. In this paper we compare performances of explicit,
implicit, and hybrid machine learning models in several publicly available
intrusion detection problems. Their applicability to mobile and cloud
computing is briefly analyzed. Machine learning methods in use areprovided by
the Weka/MOA and R/Revolution environments.
VII2.4
THE EFFECTIVNESS OF GABOR FILTER, PRINCIPLE COMPONENT
ANALYSIS AND HIGHER MOMENT ON FACE RECOGNITION SYSTEM
Salem Muftah Abdusalam, Al-Mergeb University, Al-Khums, Libya
This paper shows the effectiveness of features on the
performance of face recognition systems. The general idea is that the
features based on PCA have the greatest discriminatory value. Experiments
showed that applying Gabor Filter on ORL database and then extracting PCA and
calculating higher order moments (HiMo) combing them together improves the
performance of face recognition system. It was experimentally determined
profile of number of higher order moments in relation to the number of PCA
features for fixed dimensionality of feature vectors. Key words, PCA, Gabor
filter, Support Vector Machine.
VII2.5
MATHEMATICAL MODEL AND SIMULATION OF JAK-STAT SIGNALLING PATHWAY
Nataša A. Kablar,
Faculty of Computer Science and Lola Institute, Belgrade, Serbia
Vladimir Kvrgić, Lola Institute, Belgrade, Serbia
In this paper we give
main biological mechanism of JAK STAT
signalling pathway. We give set of biochemical reactions and mathematical
model of JAK STAT signalling pathway. We look for the control elements and
recognize SOCS1, SHP 2, and phospahatases PNP and PNX to be the main control
elements. For the chosen parameter data from the literature we run
simulations, and we give qualitative
conclusions.
VII2.6
THE APPLICATION OF THE BAYESIAN NETWORKS TO THE UNCERTAINTY
DURING THE PROJECT MANAGEMENT
Nataša Glišović, State University of Novi Pazar,
Department of Mathematical Sciences, Novi Pazar, Serbia
Risk is unavoidable in the project. The risk management has
become an important part of the project management. Although in the
literature there are a number of techniques, the risk management is rapidly
developing and managing uncertainties in complex projects is still a
challenge. There is great interest in the application of the Bayesian
networks as a tool for modeling uncertainties in project management. The
Bayesian networks provide a method for representing the relations between the
variables (Bayesian nodes in the network) even when these relations contain
uncertainty. This paper considers the problem of including uncertainties in
the project management. The proposed approach is based on the application of
the concept of Bayesian networks while modeling the uncertainty of some
activities in terms of time and cost. Validity of proposed approach is tested
on a project of franchising implementation in PE of PTT Communications
“Srbija”.
VII2.7
AN ALGORITHM FOR THE GENERATION AND EXCHANGE OF CRYPTOGRAPHIC
SECRET KEYS OVER PUBLIC CHANNELS
Sasha Adamovic, Singidunum University, Belgrade, Serbia
Milan Milosavljević, Singidunum University, Belgrade,
Serbia
This paper proposes a new algorithm to generate
high-quality cryptolographic secret keys based on shared randomness and
public discussion over public
channels. The key subsystem is correlated multivariate source of randomness,
which is based on data received from the public service monitoring globall
flight data of civilian air transport. Experimental verification of the
proposed protocol confirms its theoretical expected performance.
|