Metaheuristic Automated Design

Automated algorithms and circuits conception for signal processing.



Duration: 3 years
May 2008 to April 2011
Extention: 3 years
Launching on April 2012
Team leader: François Gagnon

Project outline

Nowadays, modern digital communication systems need to integrate an increasing variety of operation modes and modulation formats. The numerous constraints and specifications involved increase the complexity of design and hinder their development.

Paradoxically, like for other digital systems, the time allocated for the development and fabrication of communication systems either remains the same or tends to be reduced. In order to adapt design techniques to this reality, we propose to develop a new method of computer aided design (CAD). This method will encompass the formulation of multiple performance criteria and facilitate the inclusion and respect of a wide array of operation constraints. Based on utilisation of metaheuristics, the proposed method represents an alternative to the traditional design philosophy – top-down – since it considers the entire problem, without partitioning of the circuit or system functionality.

The global integration of the functionality leads to the identification of new architectures and translates in the physical realisation of digital circuits (filters, Fourier transform, equalizers, feedback loops, etc.) benefiting from better execution speeds, lower power consumption, and occupying less chip space.

The innovative nature of the proposed approach resides principally in the choice of the quantity, type, and arrangement of components: it is entirely defined by the design algorithm.

Acrobat Download the scientific overview (pdf)

Results to date

  • Development of methods and algorithms for the automated design of digital signal processing (DSP) circuits
  • Discovery of new architectures and topologies for high speed digital signal processing
  • Optimized circuits’ implementation on digital platforms (FPGA, ASIC) and comparison of their performance to more classical architectures, with respect to maximum clock speed, complexity, and power consumption

Expected results

  • Grant application is pending approval

Our partner

  • FQRNT
  • Université de Montréal

Team members

Professors: François Gagnon, Claude Thibeault and Michel Gendreau
Specialist: Pascal Giard
Post-doc researcher: Marc Joliveau
Students: Katayoon Moazzami, Tarek Baallal, François Gourlay