Projeto automático de circuitos integrados analógicos utilizando metaheurísticas bio-inspiradas

Abstract

The continuous reduction of the dimensions and supply voltages of the integrated circuits makes the optimization and manufacturing processes of the circuits more complex. The analog part of a mixed integrated circuit represents a large part of the circuit sizing effort. It is necessary to dimension each device of the circuit separately in order to reach the specifications desired for the circuit. Each component of an analog circuit, whether it is a transistor, resistor, capacitor or etc., represents one or more variables, such as the length and width of the channel of a transistor. In circuits with many variables, the combination of dimensions of the different components is very large, which makes the process of sizing and adjusting the values very complex. The designer needs to scale each of the components and check if the circuit meets the desired performance aspects. This analog integrated circuit sizing process can be modeled as an optimization problem and solved by optimization heuristics. However, the ability of heuristics to explore the search space has a great influence on the result of the circuit sizing. In this work, two bio-inspired metaheuristics were implemented to size three operational amplifiers in CMOS technology: Cuckoo Search (CS) and Firefly and compared with the results of sizing with Particle Swarm Optimization (PSO).In addition, a study was carried out on the stopping criterion for the proposed algorithms and also on the initial values for the free variables. The results show that PSO and CS are more adequate to achieve the required performance that meets the specifications of the circuit, while the FA presents difficulty in efficiently searching the design space. In addition, the results show that heuristic algorithms are suitable for the automatic sizing of analog circuits and can lead to better results than in the manual process.