Backtracking ACO for RF-Circuit Design Optimization

Abstract

Sizing analog, mixed-signal, and particularly radio-frequency circuits generally relies on the experience of the designer. Metaheuristics have been recently proposed, and it has been shown that they can arguably replace the classical iterative simulation-based trial/error approaches. Among the most known metaheuristics, Ant Colony Optimization (ACO) has already successfully been used to deal with analog circuit performance optimization. Despite the fact that ACO is robust and has a good intrinsic intensive research strategy, it suffers from its greedy requirement in computation time. In this chapter, the authors deal with the improvement of the ACO technique by integrating in this algorithm a backtracking search technique that has the role to act on the ratio of the accumulated pheromones, thus reducing the computation time. The new algorithm is called BA-ACO. Performances of BA-ACO are highlighted through some RF-applications. ADS simulation results are given to show the validity of the proposed algorithm.