Project Proposals for William Flynn Scholarship: IIndex page

Project Number: 4
Project Title: Hybrid Genetic Algorithms for Analogue Circuit Synthesis
Project Supervisor: Dr. Liam Mc Daid

Rationale for the project
Digital network synthesis has been developed to a point where most designs can be generated automatically on a computer. A specification written in a hardware description language such as VHDL can be compiled into a form suitable for programming a FPGA with little or no human intervention. In contrast, for analogue network synthesis no acceptable automatic design tools is available and consequently most analogue network design is still performed manually. Although the analogue part of most electronic systems is often a small subsection of the overall system, it is usually the most expensive because of the relatively long design time. To avoid these time to market delays, analogue design needs to become automated and to this end, attempts have been made to automate analogue circuit design methodologies using Genetic Algorithms (GA's) and Genetic Programming (GP), but with limited success relative to the domain of digital design. This is attributed to the interdependencies between component characteristics that determine the overall circuit performance. Recent work has produced circuits using a GP search technique and GA's have also been used to evolve electronic filters. The results of experiments from the above work did produced suitable circuits with appropriate outputs responses, but these filters could also be designed using conventional methods.

Project definition, aims and objectives

The project aims to investigate the performance of a hybrid intelligent GA in the domain of both passive and active analogue circuits synthesis. This work builds on previous research carried out by ISEL, but will significantly advance this research by utilising intelligent reasoning to yield an optimal analogue circuit topology from a targeted search space for a specific application. This approach, which is to the authors best knowledge is novel, aims to incorporate a reasoning capability, derived from fundamental electrical principles, human experience and the application environment, will serve to narrow the search space for the GA. Essentially what is being investigated is the ability of an intelligent GA to design "first time correct" analogue circuits, and eventually systems, for a specified application domain.

The major outcome of the project will undoubtedly be an intelligent design tool with the ability to create analogue topologies at different levels of abstraction. The project will enable an improved theoretical understanding of issues related to the application of intelligent reasoning to reduce the search space and hence the rapid identification of acceptable topologies for analogue subsystems. In addition, the research outcome from a successful completion of this project would provide valuable insight into the theory underpinning evolvable analogue hardware. There are also obvious benefits to the academic community, as the published papers will demonstrate the effective integration of intelligent reasoning and GA's and hence of the potential of a hybrid approach to problem solving in the domain of time-varying quantities. The work will also be of interest to the industrial community, as it offers the potential for rapid prototyping of analogue circuits and systems, hence accommodating the twin challenges of shorter time to market and efficient design exploration. In addition, the availability of an analogue synthesis tool would significantly advance modern mixed-signal electronics design especially for System-On-Chip solutions.

If you are interested in being considered for a studentship please contact
the Group Director, Professor T.M. McGinnity by email:
tm.mcginnity@ulst.ac.uk

or telephone: +44-(0)28-71375417.

See the current research section of this website for details on research projects pursued by existing PhD students