April 3: 8:00 – 12:00 AM
1. Principles of mobile communication viewed under a Maxwellian context: Dr. Tapan K. Sarkar
2. Neural networks and their applications to electromagnetic modeling: Dr. Christos Christodoulou
3. Diversity Combining in Fading Channels: Dr. Lal Godara
4. Dielectric resonator antenna, theory and design: Dr. Ahmed Kishk
April 3: 1:00 – 5:00 PM
5. Finite element method in time and frequency domains for solution of electromagnetic field problems: Dr. Magdalena Salazar Palma
6. Use of higher order basis in solution of electromagnetic field problems: Dr. B. Kolundzija
7. Application of genetic algorithms in electromagnetics: Dr. Randy Haupt
8. Antennas for wideband and phased array applications: Dr. Ahmed Kishk and Dr. Atef Elsherbeni
SC-1: PRINCIPLES OF MOBILE COMMUNICATION VIEWED UNDER A MAXWELLIAN CONTEXT (Tapan K. Sarkar, Syracuse University)
Nowadays, we hear frequently that we need to invent new antenna theory, or that antenna is a channel and it is something different from a conventional antenna and so on. However, from a Maxwellian point of view, antenna theory has been fixed as the northern star at least for a hundred years and even today what we called Maxwell’s equations has withstood the erosion and corrosion of progress. Even relativity had little effect as it is built in. Therefore introducing terminologies like Smart Antennas appears to lack a scientific justification. The objective of the short course is to present the Maxwellian point of view and initiate a dialog as to why the principles of mobile communication particularly dealing with MIMO and broadband channel modeling can be justified from a strictly scientific point of view. The following topics will be discussed in the half day course.
1. How does MIMO perform near field beamforming and how does it realte to ray tracing in near field analysis?
2. Are researchers modeling a perfectly dispersionless channel as we know air is diespersionless?
3. How does the Shannon Channel Capacity justify the existence of a 56 kilobit/sec modem currently installed in modern computers, communicating over a 3 kHz bandwidth telephone channel?
4. Reciprocity is a more powerful methodology than MIMO beamforming even in a complex near field electromagnetic scenario.
5. How does one accurately analyze the vector electromagnetic problem which is related to wireless than the scalar acoustic model which is not related to a wireless model?
6. Do we need to develop new communication techniques based on estimation rather than on detection?
7. Can the Sommerfeld analysis of an antenna radiating over an imperfect ground plane accurately validate the empirical Hata model and its derivatives, and accurately predict the experimental data of Okamura?
SC-2: Neural Networks and their Applications to Electromagnetic Modeling (Christos Christodoulou and Amalendu Patnaik, University of New Mexico)
Artificial neural network (ANN) methods have recently been recognized as a new alternative for RF and microwave modeling. These are now becoming an emerging tool in enhancing the effectiveness of computer-aided modeling and design of RF and microwave systems. In the last decade or so, ANNs have found several applications in the design of antennas as well. The dominating aspect in antenna technology is the search for mathematical models that will predict practical antennas more precisely and hence sharpen CAD techniques in manufacturing. ANN’s can learn and generalize from data allowing antenna and microwave circuit model development even when component formulas are unavailable. ANN models are easier to update as technology changes. ANNs are universal approximators allowing re-use of the same modeling technology for both linear and nonlinear problems. Yet, ANN models are simple and model evaluation is very fast. There is also an increased initiative in integrating of ANN technique with the existing numerical methods for the efficiency enhancement of the exiting computational models, such as, FEM, MOM, and FDTD. This tutorial will introduce the fundamentals of using ANNs for antenna analysis and design. This will also bring the participants to the forefront of this emerging field.
1. Introduction and Overview
2. Neural Network Structures
3. Training of Neural Networks
4. Modeling and Optimization for Antenna Design
a. ANN Models for Antenna Design
b. ANN Models for Antenna Analysis
5. Smart Antenna Modeling
6. Application of ANN for Computational Electromagnetics
7. Concluding Remarks and Emerging Trends
SC-3: Diversity Combining in Fading Channels (Lal Godara, The University of New South Wales)
In mobile communication channels the received signal is a combination of many components arriving from various directions due to multipath propagation resulting in a large fluctuation in the received signals. This phenomenon is called fading. In this tutorial a brief review of fading channels will be presented, distributions of the signal amplitude and the received power on an antenna will be developed, analysis of a single antenna noise limited as well as interference limited system in Rayleigh and Nakagami fading channels will be presented by deriving results for average bit error rate and outage probability. The results would show how fading affects the performance of a single antenna system. Then a comprehensive analysis of diversity combining, which is a process of combining several signals with independent fading statistics to reduce the large attenuation of desired signal in the presence of multipath, will be presented. The diversity combining schemes described and analysed will include selection combiner, switched diversity combiner, equal gain combiner, maximum ratio combiner, optimal combiner, generalized selection combiner, cascade diversity combiner and macroscopic diversity combiner. Both noise limited and interference limited systems will be analysed in various fading conditions by deriving results for average bit error rate and outage probability
SC-4: Dielectric resonator Antennas, Theory and Design (Ahmed A. Kishk, University of Mississippi)
Recently, interest in small efficient antenna has increased. One of the candidates is the dielectric resonator antenna (DRA), which is made of high dielectric constant materials and mounted on top of a ground plane or on a grounded dielectric substrate of lower permittivity. This antenna is more efficient than the microstrip antenna because of the absence of conducting edges. Also, wideband DRA are possible. The techniques used to achieve broadband DRA antennas are discussed. Some DRA’s has achieved over 50% bandwidth. The short course provides an overview for the development of the DRA. The theory and design principles and the radiation mechanisms will be discussed. Several excitation techniques are discussed with their applications to different DRA types. The DRA array design and performance are also considered.
SC-5: Finite Element Method in Time and Frequency Domains for Solution of Electromagnetic Field Problems (Magdalena Salazar Palma, Universidad Politecnica de Madrid)
The objective of this half day short course is to present the finite element method for the solution of electromagnetic field problems in the time and frequency domains. Particularly, the use of the entire domain associated Laguerre polynomials generate a time domain formulation that gets rid of the Courant stability condition and therefore generates an unconditionally stable time domain methodology without the time variable. Examples will be presented to illustrate the application of this new time domain technique. In addition, the frequency domain formulations will also be described and the use of Nedelec elements which generate a system matrix with low condition number will be outlined. Use of a higher order basis improves the arte of convergence. Finally, the use of an exact radiation condition in a finite element methodology will be presented for the solution of open region problems.
SC-6: Use of Higer Order Basis in Solution of Electromagnetic Field Problems (B. Kolundzija, University of Belgrade)
The objective of this half day short course is to illustrate the use of higher order basis functions which provides faster convergence. This is also true for finite-element and also for time domain techniques. For integral equations, it guarantees continuity of the charge. Results are presented to illustrate this point. A higher order basis function has a higher degree of continuity. For example, the pulse function is piecewise continuous function and is a polynomial of zero degree. The linear triangle function is a first order basis function as it is a polynomial of first degree. A higher order basis in this context will then deal with polynomials of degrees greater than one. We will deal with polynomials up to the ninth degree. Therefore, use of a higher order basis not only guarantees continuity of the function but also a few of its derivatives. However, we have to very careful in dealing with a higher order basis. This is because the charge is discontinuous at the feed point of an antenna and also at the end of the structure where the current with the appropriate orientation either goes to zero or has a singularity. Hence, the charge is discontinuous. We demonstrate in this course that use of higher order basis over electrically large patch sizes offer a computational advantage as the number of unknowns scales quite moderately with size and frequency. This is true not only for the solution of the integral form of Maxwell’s equations but also for the differential form. However, in using a higher order basis one has to be very careful as increasing the basis beyond a certain order may deteriorate the condition number of the matrix equation that needs to be solved for. Hence a compromise needs to be made between the choice of the order of the basis and the condition number of the matrix. It has been our experience that if the polynomials beyond the ninth order are not considered in the expansion, then the resulting matrix equations are quite stable and can be solved in an accurate fashion. Both theoretical analysis and numerical examples will be presented to illustrate these subtle features.
SC-7: Application of Genetic Algorithms in Electromagnetics (Randy L. Haupt, The Pennsylvania State University)
Numerical optimization helps us find the "best" design for a given application. Traditional optimization methods are based upon analytical formulations that were derived to find a local minimum. Most realistic designs today, however, have many local minima and a large assortment of variables. The genetic algorithm (GA) has caught on as a way to optimize practical designs. This course is intended to introduce the student to numerical optimization, GAs, and GA applications in electromagnetics problems.
1. Local optimizers and the advantages of Gas
2 Introduction to GAs – Algorithm details plus MATLAB implementation
a. Parameter selection
b. Multiple objective optimization
c. Hybrid GA
5. Applications in electromagnetics – arrays, horns, reflectors, adaptive antennas
6. GA relatives:
a. Particle swarm optimization
b. Simulated annealing c. Ant colony optimization