Design of Active-RC Filters Minimizing Sensitivities with Ant Colony Optimization

Abstract

This work employs Ant Colony Optimization (ACO) to solve active-RC bi-quadratic filter design considering passive sensitivities minimization. We propose a new discrete version (D-ACOR) based on the operating principles of ACO with real domain. We also solve the problem exactly to compare the results of the algorithms against the best possible solutions. For showing the D-ACOR performance, we take as case studies two biquadratic filters: a low-pass and a bandpass, both with an infinite-gain multiple-feedback topology. The results are excel-lent because the algorithm achieves solutions with minimal sen-sitivities that meet the specifications. Furthermore, the performance comparison with other heuristics shows that D-ACOR has lower sensitivity values and presents better computational efficiency even if it requires a more significant number of independent runs to find successful solutions.