|
|
Abstracts - Chexbres 2005
Solving Maxwell's equations using interpolating wavelets
Pedro Pinho, Paulo J. Ferreira, Sonia M. Gomes, Jose R. Pereira
View: full paper | slides
FDTD remains one of the most often used numerical methods for the simulation and analysis of electromagnetic structures. It is based on a sampled (physical) representation of the electromagnetic fields, leading to a number of advantages that include the ease with which sums, products and derivatives can be implemented, and the simple and elegant way of handling boundary conditions.
However, FDTD leads in most problems to dense (non-sparse) vectors and matrices, determined by uniform meshes that may adequately represent localised details, but that usually oversample elsewhere in space, increasing the computation time.
For this reason there is a growing demand for adaptive numerical methods based on adaptive meshes, that is, meshes that are finer in regions containing transients, and coarser elsewhere. Since fields evolve with time, spatial adaptivity is not enough. Time varying meshes, able to track the evolution of the solution in time as well as in space, are also necessary for the efficient solution of electromagnetic problems.
One way of obtaining these sparse adaptive meshes is to drop sampled (physical space) representations, and adopt wavelet-based representations. First generation wavelets lead to the required adaptive representations, but unfortunately do not seem as attractive when dealing with general boundary conditions, or when computing nonlinear functions such as the product.
The methods adopted in this work emphasise second generation interpolating wavelets. Because these wavelets interpolate, it is possible to maintain the physical space representation, now based on nonuniformly sampled fields, and to handle boundary conditions in a simple way. Furthermore, the computation of products and sums is not difficult, and approximations to the derivatives can be obtained based on finite differences.
We discretized Maxwell's equations using two meshes, corresponding to the electric and magnetic fields. The meshes can be either staggered or non-staggered, and adapted in space and in time to the nature of the fields. A comparison of the performance of the staggered and non-staggered meshes is presented, and conclusions are obtained as far as the stability and dispersion are concerned. In order to show the feasibility of the methods, 1D examples and comparisons with FDTD are shown.
An optimized spline generated horn profile with low cross-polarization
Irene Jensen
View: full paper | slides
In many applications a smooth-walled horn is preferred over a corrugated one due to manufacturing and cost arguments. Corrugated horns are known for their low cross-polarization over a broad bandwidth, while smooth-walled feed horns, like the Potter horn, have low cross polarization over a narrow bandwidth. However, by optimizing the profile of the smooth-walled horn it is possible to obtain low cross polarization over a broad bandwidth to meet specifications for most feed horns. The horn is modeled by the mode matching method, and the radiation patterns are obtained by scaling the radiation patterns for each mode by the appropriate mode conversion factor. The profile of the horn is described by the cubic spline interpolation with 7 inner and 2 end nodes. Optimization is performed with the particlesq swarm optimization (PSO) procedure, a global optimization method based on the social behavior of swarms, where a set of particles are searching the solution space for the best solution. To speed up the optimization the scattering parameters for the waveguide steps and the general radiation patterns are pre-calculated, and only the waveguide section lengths are varied to fit the spline generated profile. This method of analysis and optimization is applied to a Ku-band feed horn suited for DVB-RCS. Both theoretical and measured results are presented.
Potential antenna applications of metamaterials
Andrea Alu, Filiberto Bilotti, Nader Engheta and Lucio Vegni
View: full paper | slides
In this talk we give an overview of the potential ideas we are currently working on in the area of the application of metamaterials to antenna problems. Metamaterials are artificial materials whose electromagnetic properties may go beyond the limits that conventional media present. In particular, here we are interested in overcoming the limitations that common loading materials may show in some antenna applications. Our preliminary results show in fact that a wise employment of metamaterials with negative real parts of their electric permittivity (epsilon-negative materials), or of their magnetic permeability (µ-negative materials), and of both of them (double-negative materials), when combined together and/or with common media (double-positive materials), may present interesting properties potentially employable for new antenna setups.
Preliminary results on the effects of such metamaterials in cavity-backed reflectors, patch antennas and leaky-wave antennas are here presented, giving some insights into the reasons why such employment may be usefully applied to some realistic setups. In particular, we have designed sub-wavelength resonant antennas, low-scattering large cavities and high-scattering small cavity reflectors, and planar and cylindrical leaky-wave antennas with unconventional properties. Here we show their main radiation and matching properties, relating their unusual features with the physical properties of such metamaterials and their combination.
Dielectric inserts in layered media with equivalence principles and stratified Green's functions
Francisco Nuñez, Anja Skrivervik and Juan Mosig
View: full paper | slides
This paper presents the electromagnetic characterization of a dielectric body embedded in a multilayered structure which may contain 3D conductive structures and slots in ground planes. The approach is based on the resolution of the appropriate integral equation using the method of moments. The dielectric body is represented by equivalent currents on its surface by the means of the equivalence principle.
This paper treats in particular of the interpolation problem of the relevant multilayered Green's functions.
The 3D interpolation of electric and magnetic potentials plus the field Green's functions dyadics needed to solve the MPIE [1] functional, force us to extract all singularities from the dyadics with discontinuities in z-axis between interfaces. By cancelling their singular behaviour, a smooth function is obtained, suitable to be interpolated with enough accuracy by using linear interpolation. Two different type of interpolations are treated, one for potential Green's function and another for field Green's functions. Potential functions contain a singularity of type 1/r, where "r" is the distance observer-source, and they are not discontinuous in z-axis by using the "formulation C" presented in [2].
Otherwise the dyadic terms of the field Green's function GEM are discontinuous and a special treatment is necessary in order to interpolate them, mainly the extraction of the singular part and first image on the closest interface with change of dielectric. A simple example of a dielectric resonator notch filter is presented and compared with measurements [3].
On the effective area of antennas in phased arrays used as focal-lines feeds in cylindrical reflectors
Christophe Craeye
View: full paper | slides
A theoretical background is provided for the computation of the effective area of finite-by-infinite arrays and it is applied to the case of cylindrical reflectors fed with complex focal-line arrays. Results based on full-wave Method-of-Moments are shown. The simulations involve both the phased array and the reflector.
Mutual coupling correction in antenna arrays for DOA estimation
Lajos Nagy and Akos Ferenc Nemeth
View: full paper | slides
The mutual coupling between adaptive antenna array elements degrades the performance of the array especially for direction finding purpose. There are many possibile ways to compensate the coupling effects via electromagnetic analysis or via signal processing. We are following the first approach, calculating the mutual coupling matrix and compensate its effects. The mutual coupling is investigated for wire and for microstrip structures.
The simulation for the arrays especially for the 4 by 4 array is based on the element by element analysis because of the large structures.
The wire structure is analysed using a MoM program developed at our department. It is based on the Galerkin solution of the Pocklington integral equation using picewise linear test and weighting functions.
The microstrip structure is analysed using Ansoft Designer and a FEM-MoM program developed at our department. The FEM solves the weak form of the three dimensional integral equation using tetrahedron test functions and again Galerkin approach is used. On the border of the volume analysed a PML boundary condition is used.
In the paper the effect of coupling on the DOA estimation and the corrected results are also introduced.
Adaptive radio direction finding measurement results and the aspects of measurements, investigated with a re-configurable, 16-channel radio direction finder antenna array
Gyula Miko, Andras Magyar, Rudolf Seller
View: full paper | slides
A recently developed re-configurable 16-channel radio direction finder antenna array is presented with measurement results. The array has 16 I-Q channels and it is capable of measuring the direction of radio sources radiating at 4500 MHz in 2D. First, few words are said about adaptive radio direction estimation in general, then the structure of the system is explained. Following this, usage aspects, such as calibration, sampling dependency of adaptive algorithms, EM environment, regarding the presence of multiple correlated sources, and array structure, effecting measurement results are discussed. Our measurements were conducted in a radio-anechoic chamber. Adaptive methods were also investigated and are discussed in the paper.
Tracking of a geostationary earth orbiter using an electronically steerable inexpensive earth station
Patrice Siatchoua and Roman Gieron, IMST GmbH
View: full paper | slides
Satellite communication from mobile vehicles needs an automatic tracking of the antenna pattern in order to maintain the connection to the satellite. In most of today's terminals the problem has been solved using mechanical steering: The antenna is mounted on a two axis steering system, that adjusts the antenna orientation in such a way, that the main beam is continuously re-directed towards the satellite. However, mechanically moving parts are sensitive and have limited lifetime, so there is an increasing demand for electronically steerable solutions.
Within the INES (INexpensive Earth Station) project, a conformal, electronically steerable and automatically tracking antenna for future satellite communications is being realized. The scope of this activity is to develop a solution for land mobile applications (cars, trains, trucks) in L-band. In order to reduce complexity and costs, the antenna is based on a combination of phased array and beam switching concept.
During tracking, the relative position of the satellite is estimated based on data from inertial measurement unit, compass and GPS receiver. This paper presents some design results of the INES antenna and describes a method of applying the techniques of data fusion to the design of an antenna system which can automatically track GEO satellites.
Mutual coupling analysis of cylindrical waveguide arrays using hybrid SD-UTD method
Zvonimir Sipus, Marko Bosiljevac and Sinisa Skokic
View: full paper | slides
One way of analyzing conformal arrays is to use the separation of variables and express the solution in a modal or harmonic series expression. Such a solution can be obtained only for some special cases, like a perfectly conducting circular cylinder, elliptic cylinder, sphere or cone. The separation of variables method can be generalized to include an arbitrary number of dielectric layers. For this application, it is convenient to apply the two-dimensional (2D) Fourier transformation in the coordinates for which the structure is homogeneous (for example, in the cylindrical case we perform the Fourier transformation in axial direction and the Fourier series in phi direction). For each spectral component of the source, the excited electromagnetic field in the two directions for which the structure is homogeneous has the same harmonic variation as the source. Therefore, the original three-dimensional problem is transformed into a spectrum of simpler onedimensional problems. Since each one-dimensional problem is solved in the spectral domain, this approach is also called the spectral domain (SD) method.
The two biggest disadvantages of the spectral approach are in summation of high number of terms in the Fourier series for structures with large radius, and in evaluating highly oscillating integrals when the source and observation points are separated in the direction for which the structure is planar (e.g. in axial direction for cylindrical case). There are a lot of techniques to accelerate the inverse Fourier transformation/series. In the presentation we will describe the hybrid Spectral Domain - Uniform Theory of Diffraction (SD-UTD) technique for reducing the needed number of terms in the Fourier series and the needed length of integration in the Fourier transformation. The basic idea is to subtract the asymptotic part of the Green's function, and to calculate the asymptotic part using Uniform Theory of Diffraction (UTD). The properties of the method will be demonstrated on analyzing a waveguide array embedded in multilayer cylindrical structure.
A cylindrically shaped antenna subarray generating four beams
Pawel Kabacik, Krzysztof Wincza, Slawomir Gruszczynski and Krzysztof Sachse
View: full paper | slides
In this presentation, we describe part of our research which focused on techniques for generating multiple beams with cylindrical arrays, and on creating relevant beamforming networks capable of operating over a wide band. It was recently reported that providing a multibeam radiation pattern is possible with a passive circuit manufactured at low cost [1].
For such types of beamforming network, we designed and manufactured models of the broadband Butler matrix. The directional couplers used in our matrix may be of two-strip or three-strip design. These couplers feature broadband properties and have low losses. Both designs make use of transmission lines placed on the two sides of a thin microwave laminate (5 mils or less). The design of the couplers was done with software developed in-house and was verified with Microwave Office, a commercial CAD tool. The measured amplitude and phase transmission characteristics feature remarkably small phase variations, with the phase characteristic varying within ±1 degree across the 1000 MHz band centered on 2400 MHz. The amplitude balance between the four output ports is narrowly kept, and the maximum differences approach 0.5 dB. It is worth mentioning that the simulated and measured results were highly consistent each other. To provide an arbitrary value of a phase shift in some instances (e.g. 8x8 Butler matrices), Schiffman shifters were examined. The upgrade of the Butler matrix to the 8x8 format raises problems with line crossings, particularly with the stripline concept.
The presented broadband feed was implemented in a cylindrical subarray that we developed in-house. The cylindrical subarray is manufactured with the antenna ground of 195 mm radius. The subarray was made as a facet panel of four elements featuring four beams generated with a Butler matrix. The measured shapes of the four beams are consistent with the calculated ones. The simulated data was obtained with the translation and rotation method, which we implemented in a computer code. The pattern is not precisely optimized against the smooth slopes of the main beams (shoulder-like distortions were avoided with inserted sections of lines). Further study is needed to find a method of suppressing the sidelobe levels and improving polarization properties in such a multiple beam conformal array.
When the 1000 MHz band is available, the developed broadband Butler matrix is capable of serving numerous mobile and wireless systems allocated from 1700 to 2500 MHz. With such broadband feeding network, patch elements featuring dual band or wideband operation should primarily be used. Arranging the elements in the cylindrical array results in their strong interactions and thus special attention must be paid to keeping the side effects due to coupling under designer control.
Estimation of signal correlation between antennas from their radiation patterns
Jussi Rahola
View: full paper | slides
For diversity applications, the correlation between antenna signals is a key parameter for the evaluation of the diversity performance of the antennas. With certain assumptions of the radio channel, the correlation can be estimated from the radiation patterns of the antennas. The widely used correlation formula assumes that the vertical and horizontal components of the incoming radiation in the radio channel are independent. A new formula is derived that includes also the correlation between the two components of the radio channel. This formula allows the modeling of radio environments where the incoming signal is, e.g., circularly polarized or has a slanted 45 degree linear polarization.
Characterization of mobile phones on reception by measurements of total isotropic sensitivity and average fading sensitivity in reverberation chamber
Charlie Orlenius and Per-Simon Kildal
View: full paper | slides
The reverberation chamber, or mode-stirred chamber, is well-known and widely used in the area of electromagnetic compatibility (EMC) testing to create a statistically repeatable field distribution. The basic theories are well investigated and described in journal articles.
The antenna group at Chalmers University of Technology together with the spin-off company Bluetest AB have been dealing with the issue of measuring the performance of antennas in a relatively small reverberation chamber. The smallest dimension of the chamber is in fact approximately three wavelengths of the lowest frequency. Several improvements have been introduced to get the desired accuracy. These developments are described in several papers and reports. We have also derived procedures to measure radiated power of active terminals, as well as diversity gain. In the present paper we will describe two ways to measure receive performance of active terminals by using a reverberation chamber, referred to as total isotropic sensitivity and average fading sensitivity. The latter is a new method which can be measured very fast in a reverberation chamber. It is also more representative of the actual performance of a terminal in a real fading environment than the former.
Measured radiation properties of a broadband terminal antenna
Pawel Kabacik, Robert Hossa, Arkadiusz Byndas, Per-Simon Kildal and Sadullah Özcan
View: full paper | slides
The emergence of a wide range of applications for small antennas has produced significant works in the area of miniature antenna measurements. This presentation is concentrated on measurement methodologies applied to test broadband small antennas for use in modern terminals. One method developed at Chalmers is capable to measure a number of key parameters, among them the most important are total radiated power, radiation efficiency and diversity gain. Another method developed in-house at Wroclaw University is for measuring radiation patterns of small antennas.
Optimization techniques to reduce focusing element abberations
E. Jehamy, G. Landrac and Michel M. Ney
View: full paper | slides
The LEST is presently carrying out some research in developing alternate three-dimensional antennas, which are implemented in foam technology. The foam advantage is the potential low-cost and flexibility as it can be arbitrarily machined within few micron accuracy. In addition, foam has attractive electromagnetic properties such as relative permittivity close to unity, low loss up to 110 GHz and can be metalised by simply spray-painting or hot pressing and gluing. The presentation will focus on reflector and constrained metal plate lenses (artificial lens) which are currently under investigation at the LEST for automotive radar applications at 77 GHz. It is generally required that the gain be above 25 dB with side-lobe levels below -20 dB. The lower the side lobe level the lower the false target detection rate of the radar system for a given signal processing algorithm complexity.
As the antenna should provide some beam control feature, the primary source position of the focusing element should be moved off focus. As a result, aberrations will occur, producing both gain reduction and higher side lobe levels. A general procedure to reduce these aberrations is presented. The objective is to design such element with minimum phase errors (aberrations) to reduce side lobe level, for relatively large beam tilt angles. The procedure is described in details in the case of various configurations. As dimensions involved are of (12 lambda)^3 order, full-wave analysis is precluded for fast design. Based on a simple Geometrical Optics (GO) computations, various optimisation algorithm such as variational method (VM) or Genetic Algorithm (GA) are used to find the lens or reflector profiles. In addition, some optimisation using a MoM full-wave analysis is extended to the primary source to further improve the results. Finally, a full-wave analysis is performed to assess le validity of the procedure. Various examples of structures will be shown.
Gaussian beam antennas based on open resonator structures
Giuseppe Di Massa, Luigi Boccia and Giandomenico Amendola
View: full paper | slides
Millimeter waves frequency bands are very attractive resources for telecommunications as they can be used in many applications which includes the realization of high data rate links in pico cellular networks, local multipoint data services, automotive radars, inter-satellite communications etc. The design of antennas at such high frequencies cannot be afforded with conventional designs. In fact many original solutions like printed antennas on membrane structures, reflectarrays and many kind of lens antennas have been proposed. Gaussian beam antennas are a very interesting solution as they can provide very low multipath thanks to the low sidelobes level. They are based on the known result that the field map at the mid section of an open resonator shows a Gaussian distribution that can be used to illuminate a metallic grid. In this paper a novel Gaussian beam antenna, based on an open resonator with spherical mirrors, is presented. In the following a simple circuit model based on the modal analysis of the open resonator (O. M. Bucci, G. Di Massa, "Open Resonators Powered by a Rectangular Waveguide", IEE Proc. H, vol. 139, Aug. 1992) will be given. As preliminary result a test design will be considered and the input impedance predicted by the equivalent circuit will be compared with full wave simulations. Furthermore results on the simulated gain and radiation patterns will be shown.
Multi-beam antennas employing the shared aperture concept
Cristian I. Coman, Ioan E. Lager and Leonardus P. Ligthart
View: full paper | slides
The concept of shared aperture antenna, involving interwoven sparse arrays that share a common aperture, provides a proper utensil for supporting applications that require multiple beams. Recently, an intense research activity was carried out at the International Research Centre for Telecommunications-transmission and Radar (IRCTR) for developing design strategies that aim at obtaining large, interwoven sparse arrays. The objective is twofold: on one hand, a wide-band antenna can be synthesised by simultaneously employing several sparse arrays consisting of moderate bandwidth elementary radiators; on the other hand, multi-beam applications can be supported by employing each (sparse) sub-array for generating individual, independent beams. The two targets are amenable to a unitary design strategy. It is based on the use of a deterministic deployment procedure, centred on the application of the difference sets theory. The strategy yields densely populated, shared aperture antennas, consisting of a very large number of individual radiators. By making the sub-arrays sparse, the use of amplitude tapering becomes superfluous. The deployment strategy being deterministic, a very limited computation time is required for the design of a complete configuration.
The proposed design strategy is illustrated by means of two shared aperture antennas:
1. An antenna consisting of two types of cavity-backed, patch antennas operating in neighbouring, partially overlapping, spectra in the X-band. A full-wave analysis that accounts for the mutual coupling in a non-periodic environment, as well, is employed for assessing the radiation properties of the proposed structure.
2. A re-configurable antenna consisting of 8 sub-arrays, consisting of ideal radiators and generating up to 8 independent beams of controlled beam-width.
Interference mitigation using switched beam antennas
Juliane Iten Simoes, Ivica Stevanovic and Anja K. Skrivervik
View: full paper | slides
This paper presents the results of Monte-Carlo simulations of a GSM/DCS base station site with standard three-sector directional antennas and the same base station with four-switched beam antennas per base station sector. The simulations show significant decrease in probability of interference in victim receivers at base station with four-switched beam antennas as compared to the standard three-sector directional antennas. Two models of the four-switched beam base station antenna are presented, and these models can be used to simulate a general scenario.
Contoured and/or multi beam reflectarrays design
O. M. Bucci, A. Capozzoli and G. D'Elia
View: full paper | slides
Even if reflectarrays have been introduced since many years, printed reflectarrays, due to the low costs and weight and to their compact dimensions, can now play a relevant role in modern telecommunications applications.
As a matter of fact, due to the high number of parameters (degrees of freedom) that, at least in principle, can be controlled, in the framework of multibeam and contoured beam antenna systems reflectarrays could also outperform reflector antennas.
Obviously, to this end three main points must be considered:
* an accurate electromagnetic model of the antenna should be available in order to describe both the patch reflection matrix and the diffraction effects due to the abrupt patch size variations across the antenna, the truncation as well as the surface wave effects;
* an effective, flexible and efficient synthesis technique should be available in order to fully exploit the available degrees of freedom in any practical geometry and for any far field pattern specification of interest, avoiding the trapping problems typical of any non linear optimization technique;
* when the bandwidth is of concern, new patch configurations must be introduced in order to enlarge the reflectarray bandwidth.
A new approach to the design of printed power pattern shaped beam reflectarrays based on an accurate electromagnetic model of the radiating structure and exploiting a synthesis technique based on the generalized projection optimizer has been recently developped at the University of Naples Federico II. The technique allows to improve the results achievable by other available approaches. In particular, the first worked examples confirm the feasibility of high performance contoured beam reflectarray antennas.
A simple way of obtaining optimized patterns using the Woodward-Lawson method
J. Fondevila Gomez, J. A. Rodriguez, F. Ares and E. Moreno
View: full paper | slides
Multi-beam antenna arrays have applications in the field of electronic countermeasures, in satellite communications, and in adaptive nulling. For these systems, the most natural method for synthesizing multiple beams is the Woodward-Lawson method. In this communication we describe a new way of applying the Woodward-Lawson method in order to avoid its limitations concerning pattern performance. The proposed method obtains the Woodward-Lawson beam coefficients by sampling a previously synthesized optimal pattern using the Orchard-Elliott method. The combination of both synthesis methods will provide the advantages of each: on the one hand, power patterns with a controlled ripple and sidelobe topography and also a set of coefficients of the orthogonal beams which can be realized in real life with an array feed consisting of a Butler matrix. This work also discusses new insights about the multiplicity of solutions in the Woodward-Lawson method.
Pattern synthesis by using time-modulated linear arrays
J. Fondevila-Gomez , J.C. Bregains, G. Franceschetti and F. Ares
View: full paper | slides
Time-modulation applied to linear arrays is a technique that was born in the middle of the 20th century. It consists in using on-off switching devices in the feeding network of an antenna array in order to take control on its radiated density power, by means of time-periodic rectangular pulses applied to some of the elements of such an array. The main problem of this technique is the appearance of unwanted harmonics (usually called sideband level) that have to be minimized. This idea was forsaken for many years but, recently, new results were obtained by authors as Shiwen Yang et al. and Fondevila et al., in works where optimisation techniques, such as the Differential Evolution Algorithm or the Simulated Annealing, were used.
In our research group, we have synthesized sum patterns by applying this technique to a linear array of 30 equispaced isotropic elements, working at a single frequency f. In a second approach, a linear array of collinear dipoles working at two frequencies f and 2f, was synthesized by optimizing the duration of the rectangular pulses and the position of the elements of the array as well. Finally, by considering some advantages of the technique - which will be discussed during the symposium - further applications were performed, one of them being the synthesis of difference and improved sum patterns by time-modulating only a few elements of the array fed by a uniform excitation distribution.
Further details about time-modulation applied to linear arrays such as signal transmission and power losses due to sideband radiation, will be discussed during the presentation of this communication.
Is fully controllable conformal (multi-beam or adaptive) transmit/receive radio-systems possible ?
Leo Vaskelainen
View: full paper | slides
In VTT Information Technology we have done some work for building a multi-channel digital receiving system, basically an experimental adaptive antenna solution in mind. This kind of solution is also suitable for conformal receiving antennas, because both amplitudes and phases are easily controlled by software. Multiple beams can be formed by parallel calculation.
A more interesting question is, is it possible to build an equivalent software controlled transmitting system ? Basically, with ideal components, it is possible. An idea for that kind of system is presented.
INVITED & LOCAL PRESENTATIONS
Antenna research activities at Department of Electromagnetic Field CTU in Prague
Milos Mazanek and Pavel Hazdra
View: full paper | slides
Outline of the presentation:
1. Short general overview (position within CTU, topics, projects, personel, labs)
2. Antenna group
a. history
b. infrastructure - laboratories - possibility of measurement (impedance,
radiation pattern, polarization measurement, far-field, near field, frequency
domain, time domain ...)
c. infrastructure - software tools
d. projects and obtained results
i. broadband antennas - ridged, spiral
ii. multiband antennas - fractals
iii. antennas for special measurement and testing (Nation Security Authority)
e. future plans
3. Conclusion
Antenna development at RF-spin in Prague
Milos Mazanek, Zdenek Hradecky and Pavel Pechac
View: full paper | slides
Outline of the presentation:
1. History, ideas, personel
2. Department of Electromagnetic Field and RF-spin relationship, other collaboration
3. Development and production
4. Conclusion
Broadband modeling and optimization of EM structures
Zbynek Raida and Zbynek Lukes
View: full paper | slides
Outline of the presentation:
1. Brno University of Technology: general information
2. Laboratory of Numeric Techniques
2.1 Today's research: time-domain analysis and multi-
objective optimization
2.2 Results obtained
2.3 Background: infrastructure, financial sources
3. Conclusions
Microwaves and electromagnetics at University of Belgrade
Antonije Djordjevic
View: full paper | slides
The tentative contents of the presentation is:
* About the Microwave/EM/Antenna group
* Overview of activities (teaching, research, software, projects)
* Selected projects related with the COST284 scope
Research & development activities in antennas at CICESE
José Luis Medina Monroy, Ricardo Chavez Perez, Arturo Velazquez Ventura and David Covarrubias Rosales
CICESE, Ensenada, Mexico
View: full paper | slides
FEST: A full-wave electromagnetic simulation tool for antenna waveguide networks
M. Mattes, C. Vicente, H. L. Hartnagel, D. Raboso and J. R. Mosig
View: full paper | slides
Despite the big market and wide range of applications of printed circuits, waveguides are still key components in many micro- and millimetre-wave systems like filters, multiplexers, or feeding networks of antenna arrays.
In applications of limited power supply, as it is the case in satellite systems, low-loss components are mandatory. Although hollow waveguides are disadvantageous with respect to weight, they outperform printed circuits in terms of losses which justifies their application in the space segment. But also in the ground segment they are omnipresent, especially in high power applications.
Even if the principles of using waveguides as high-frequency systems were described just after the second world war and even before, the complete analysis of a complex waveguide network, as currently found in modern telecommunication systems, including manifolds such as multiplexers and a higher number of channels, is still today a challenging problem.
Although commercial purpose software (for instance CST Microwave Studio, HFSS) is available with a high flexibility in modelling electromagnetic components, their application is limited in terms of complexity since the numerical analysis of the structure under test is based on discretizing the complete structure. Therefore, with increasing complexity the computational resources are rapidly exhausted, and for instance a full multiplexer with connected filters is beyond their capability.
Moreover, modern microwave and antenna systems also demand a higher component integration as well as increasing power handling capabilities. Both requirements lead to higher electromagnetic field densities inside the devices which results in a higher risk microwave breakdown. In fact, highly time consuming and expensive test campaigns must be undertaken since no reliable software tools exist that is able to predict the microwave breakdown power for arbitrary geometries.
Today, the time to market of a product has to be more and more shortened in order to reduce the costs. This demands for CAD tools in the design of electromagnetic components that, on the one hand, can predict the electromagnetic behaviour accurately enough. On the other hand, they also must take into account the common trend in communication systems toward higher frequencies, higher component integration and higher field densities.
Improving the effciency of the subdomain multilevel approach
Ivica Stevanovic and Juan R. Mosig
View: full paper | slides
Solving electromagnetically large structures with an Integral Equation/Method of Moments (IE-MoM) technique using subsectional basis functions is a very demanding procedure in terms of both computer memory and time. The memory needed for solving a problem of N unknowns increases with O(N2) and the complexity with O(N3) if the resulting MoM linear system is solved with a direct method (standard Gaussian elimination or equivalent). Computationally efficient techniques are thus needed to accelerate the IE-MoM procedures and allow modeling of large circuits and antennas on standard desktop PCs. There are a number of techniques used to accelerate the MoM calculations and improve the O(N2) and O(N3) factors. The fast multiple method (FMM), the multilevel fast multiple algorithm (MLFMA), the impedance matrix localization (IML), the adaptive integral method (AIM), and the multilevel matrix decomposition algorithm (MLMDA) are all iterative techniques keeping the same number of unknowns but using very efficient matrix-vector product schemes. Another large group of approaches is based on the size-reduction of the matrix and nonexhaustive list includes the diakoptics-based multilevel moments method (MMM), the synthetic basis function (SBF), the characteristic basis function (CBF), and the sub-entire-domain (SED) basis function methods.
In this letter we present the further improvements in the subdomain multilevel approach (SMA) with macro-basis functions (MBF), a technique that belongs to the latter group and that has proven to be very efficient in modeling the large printed antenna arrays. Using the SMA with MBFs, the size of the MoM matrix is significantly reduced. This leads to a sizeable drop in the CPU time needed to solve the problem, which becomes comparable to or smaller than the MoM fill-in time. We present two strategies that improve the MoM matrix filling. The first way takes advantage of the subdomains that are translated or rotated replicas of an already computed subdomain. Using the principles of translational or rotational symmetries of the subdomains in which the structure is divided, we do not need to re-compute MoM submatrices for every subdomain but rather copy the corresponding submatrix. All this presumes, however, that the subdomains are not only geometrically equal, but that they have exactly the same mesh and exactly the same basis function numbering scheme. The strategy for improving the subdomain mutual-interaction filling time is based on reducing the MBFs deÞned over subdomains to their equivalent moments. Instead of computing the mutual interactions between every pair of subsectional basis functions belonging to two different subdomains and then summing them up, the mutual interaction between two subdomains is computed as the sum of a significantly lower number of equivalent-moment interactions. In this paper we describe these improvements and demonstrate their capabilities on several practical examples.
|