Printer-Friendly Version

Background

The MTT Microwave Field Theory Technical Committee (MTT-15) provides leadership and support to the MTT community regarding all aspects of electromagnetic field theory and computational electromagnetics. MTT-15 thus has a fundamental and empowering role in promoting the theoretical and computational foundations of all microwave technologies and applications.

MTT-15 is pleased to sponsor a student design contest on "Innovative Modeling Techniques for Microwave Engineering" at the IEEE International Microwave Symposium to be held in June 2011 in Baltimore, Maryland. This competition is open to all IEEE MTT-S student members. The objective of this competition is to foster development and innovation in the following areas:

1. Methodologies and computational algorithms for enhancing the analysis and design of microwave circuits and devices
2. Efficient computing paradigms (including parallel processing and other approaches) for implementing computational electromagnetic algorithms such as the FEM, MoM, FDTD, and TLM methods.

Projects should address the specific canonical analysis/design problem detailed below.

Rules and Regulations

Design Proposals describing the proposed analysis and design for the problem are due by March 1, 2011. Teams of up to four student members are allowed, provided each team member has a direct contribution to the design project. Notifications of acceptance will be sent out by March 15, 2011. The final submission of the team project material must be sent to the MTT-15 design contest coordinator by May 15, 2011. All contestants are expected to attend IMS2011 and participate in the MTT-15 Student Competition Session, where they will give a poster (open forum) presentation to the judges and interact with them.

Requirements

All competition material submitted to the competition will be made available to the MTT community via the MTT-15 website. The material submitted for the final project evaluation must contain the following:

1. A formal report addressing the following areas:
• Problem description
• Design methodology and innovation
• Results and conclusions
2. An electronic version of the poster to be presented at the MTT-15 Student Competition.
3. An electronic version of all software packages developed.

All CAD projects/files and user-developed software modules must be submitted for evaluation. These should be documented as professionally as possible.

Scoring

The winning project will be selected using the following criteria:

1. 20% — Difficulty of the Problem
2. 20% — Modeling and Design Methodology
3. 20% — Innovativeness
4. 20% — Quality of Documentation
5. 20% — Quality of Presentation

Criteria 1 to 4 will be evaluated in advance. The quality of presentation will be judged at the conference (open-forum format).

Prize

The winner will receive a prize of US $1,000.

 

Innovative Modeling Techniques for Microwave Engineering

— Efficient computing paradigms

Computer technologies have advanced rapidly in the last decade; a normal desk side workstation now has processing power beyond most supercomputers that were available a decade ago. However, most computational electromagnetic programs cannot run much faster than their previous versions that were available ten years ago. This is because the increases in processing power comes from new computer architectures, multi-core central processing units (CPUs) coupled with massively parallel graphics processing units (GPUs), rather than from higher CPU clock rate. In fact, most CPUs are running at clock rates lower than the 3GHz limit that was achieved at the beginning of year 2000.

It is thus necessary to redesign some, if not all, popular computational electromagnetic algorithms so that the new algorithms can take full advantage of heterogeneous computing hardware. The objective of this project is to foster the design and development of massively parallel computational electromagnetic algorithms for heterogeneous computing hardware comprising CPUs and GPUs.

Problem Description

Write a CPU based field solver in C++ or Matlab; the solver should be based on a "brute-force" method such as MoM, FEM, FDTD, or TLM. Convert the program to run on a GPU based computer that supports OpenCL 1.0. Use both programs to obtain the first fifteen cut-off frequencies (not counting the dominant quasi-TEM mode) of the structure shown in the figure below. Document the algorithms, program designs, and any problems encountered when porting the field solver to the GPU environment. Obtain the required cut-off frequencies and comment on the sources of error. Validate the obtained result with a commercial software package of your choice. Discuss the advantages and disadvantages of GPU based algorithms. Draw your conclusions.

Figure 1: The cross section of an infinitely long coaxial waveguide.