InterJournal Complex Systems, 311
Status: Submitted
Manuscript Number: [311]
Submission Date: 991009
An intelligent Framework for Designing Analog Circuits Based on Hybrid Reasoning
Author(s): Farhad Razaghian

Subject(s): CX.66

Category: Brief Article

Abstract:

Design of analog integrated circuits is usually a time consuming process due to the complex relations between the design parameters and circuit performances . The best approach to design in the shortest possible time is to develop CAD-oriented tools that can automatically design analog circuits. Analog circuit design is accomplished in two steps: in the first step, the designer selects an appropriate circuit topology out of the various possible topologies or architectures, to achieve a higher performance for a particular application. The second step is based on assigning values to circuit parameters so that the circuit performances can satisfy a set of specifications. In this paper, we propose a new approach to design analog integrated circuits based on the hybridization of three kinds of reasoning: fuzzy rule-based, case-based and qualitative reasoning. Although, the topology or architecture selection depends ordinarily on designerís experience, it can however be automated with the aid of the artificial intelligence methods . It has been demonstrated that fuzzy logic, despite itís intrinsic inexactitudeness, can act as a precise method for optimal decisions, out of which design problem solving is also noticeable . The method proposed in this paper, uses fuzzy rule-based reasoning to select an appropriate architecture out of the alternatives accumulated in a library. This guaranties the best possible initial guess, and reduces the number of epochs for circuit simulation . Case-based reasoning has originally emerged as a human memory and reminding . We have used this approach to translate the input specifications into some building blocks subspecifications and to determine subsequently the circuit device sizes efficiently. The designed circuit is then simulated to show how efficiently it performs. If we come to the conclusion that some input specifications have not been met yet, a qualitative reasoning based approach , which is responsible for applying the most suitable (plausible trade-offs) variations to both circuit topology and device sizes would subsequently be used to make the final correction on the circuit topology and the related device sizes.

Submit referee report/comment


Public Comments: