Project 3-4: Antenna Front-End and Receiver Technology for Multi-Function Radio Frequency Architectures

Principal Investigators: Gregory Wilkins, Lawrence Walker (Morgan State University)

The computations necessary to design and build complex systems of antennas, which can contain millions of components, require an equally complex system of algorithms. The the behavior of the electromagnetic field within each component must be analyzed while taking into account the effects due to materials and other parameters that support and connect the components to one another.

Especially in the high-frequency regime of microwave and millimeter wave integrated circuits, the sheer numbers of components and the mesh resolution required for analysis makes high performance computing an essential tool for solving design problems and identifying and addressing factors that can limit the performance of these networks.

Gregory Wilkins, professor of electrical and computer engineering (ECE) at Morgan State University, and Charbel Farhat, professor of mechanical engineering at Stanford University, are developing fast, scalable parallel iterative solvers for large-scale electromagnetic calculations, with the assistance of Morgan State ECE students Razid Ahmad (undergraduate) and Babafemi Talabi (graduate). They will apply these problem-solving algorithms to the innovative design of antenna systems with millions of components and to creating computer simulations to assist in optimizing these designs. Shauna K. Henson (2006 ECE graduate) is working as operations coordinator for various aspects of the project, assisting with both research and outreach.

The Project to Date
The essential parts of the computing system are under development. Ahmad has been instrumental in configuring the computer network and investigating additional hardware components, such as servers. Licensed software will be loaded onto the system to allow multiple users to gain remote access to the centrally-located software tools necessary to complete the research effort. These tools will be essential for future computational electromagnetic simulations and will form the basis for the high-performance electromagnetic research program.

The group is working to develop, then verify and benchmark, a FETI-DPM (finite element tearing and interconnecting–dual/primal Maxwell) solver for Maxwell problems. (Maxwell’s equations describe the behavior of electrical and magnetic fields, such as those produced by antennas.) The solver will include modifications that allow for rapid changes, or sweeps, in the frequencies of operation of the antenna. In addition, the algorithm will be generic, allowing for multiple interactions in regions of varying material parameters, otherwise known as heterogeneous media. The algorithms will be developed and written in the form of a stand-alone, portable, massively-parallel code of the type used by supercomputers and high-performance computing clusters. The plan is for this stand-alone module to be fully integrated into Morgan State University’s computational electromagnetics (CEM) code, which will be delivered to ARL. In the longer term, a parallel AMR (adaptive mesh refinement) capability will be integrated into the CEM codes and optimally connected to the FETI-DPM solver. The resulting package promises to be among the fastest finite element modeling-based CEM solvers. (See the sidebar for a description of some concepts and methods discussed in this article.)

The Morgan State group plans to collaborate with ARL scientists to use computer simulations in the design and optimization of a series of antenna systems with millions of components, thus demonstrating the problem-solving potential of the CEM code.

Wilkins’ research group is implementing previously-investigated techniques that employ a hybrid edge element approach, which is used to represent all electric field components in the region of interest in guided-wave structures. The group intends to determine the corresponding electromagnetic field behavior for propagating and decaying modes in waveguides with irregular cross sections, for which analysis with standard techniques proves to be extremely difficult. An algorithm will be developed and implemented to account for the field behavior in these regions by considerations of the variation in both location and direction of the materials within the waveguide. Additional considerations will include the modification of the mesh element type for regions where extremely thin layers are used, as well as modifications in the vicinity of a perfect electric conductor.

Talabi and Wilkins have started an investigation of the behavior of internal and external electromagnetic fields for a variety of configurations, most generally known as the generic prolate spheroid. The focus of this research is the derivation of a general solution in closed form for this type of configuration with the long axis parallel to the electric field vector of an incident plane-polarized electromagnetic wave. The prolate spheroid may be considered with a range of essential shapes, including a fiber cylinder, for which the aspect ratio (length versus diameter) is extremely high. (At the other extreme is the oblate spheroid, often referred to as a “flat pancake.” In the middle of these two extremes is the spherically-shaped particle.)

As an initial assumption, frequencies corresponding to wavelengths much greater than the dimensions of the particle are being considered for the fiber cylinder, with an update of the solution to wavelengths on the order of the dimensions of the particle. Ideally, the frequency range will be from the megahertz to infrared ranges (“DC to daylight”). Solutions are being generated for the internal and external electromagnetic fields that are written in terms of the physical variables of the problem (particle conductivity, permeability, and dimensional parameters, and the frequency and polarization of the incident wave). This type of research is beneficial in that it allows for the study of a wide variety of antenna configurations with irregular shapes, which may be grouped together for beam steering purposes (referred to as phased arrays).

Ongoing simulation work includes the application of Ansoft’s High Frequency Structure Simulator to a Rotman Lens beamforming network. The continuation of this project will employ domain decomposition as an added feature. The results using domain decomposition will be compared with the results of simulations that do not use the added feature.

Army and Civilian Applications
In conjunction with the ARL Adelphi Sensors and Electron Devices Directorate, students are being recruited to work on the design of antenna architectures suitable for conformal application to Army platforms in support of system requirements, including receive-only broadband antenna systems and antenna arrays.

This research supports the investigation of factors that can limit the performance of very-large-scale-integrated antenna networks. Sensor networks incorporated into spaces such as homes, cars, stores, roads, or cities must take into account mobile users distributed in the same space as the network. Other factors that must be addressed include rapid information aging, cross-interference from the concurrent operation of multiple networks (cellular, WiFi, WiMax), and serious security and privacy concerns. By developing the tools and methods to assess such situations, this HPC project can identify and optimize the factors necessary to construct and maintain these complex networks in a variety of environments.

Source: AHPCRC Bulletin, Vol. 1, Issue 2 (2008)