Certified Computation of Optimal Multiband Filtering Functions
Résumé
In this paper, we focus on the problem of computing multiband filtering characteristics with a guarantee on their global optimality with respect to a Zolotarev-like criterion. An iterative algorithm based on linear programming is presented. This algorithm ensures quadratic convergence to the optimal solution. We also provide an equiripple-like criterion that allows one to check in a very simple manner whether a computed filtering function is optimal or not. The latter is used to analyze two practical design examples based on asymmetric dual-band specifications. In particular, it is shown that the selectivity of the dual-band response does not necessarily increase with the filter's order. This study yields some striking results when compared to the usual single-band situation, and introduces the idea that for certain asymmetric specifications some of the filter order values are more suitable than others. Finally, the practical implementations of the filtering devices in inline dual-mode cavities and stacked single-mode cavities are detailed.
Mots clés
Zolotarev-like criterion
asymmetric dual-band specification
equiripple-like criterion
inline dual-mode cavity
iterative algorithm
linear programming
microwave filters
optimal multiband filtering function computation
quadratic convergence
stacked single-mode cavity
band-pass filters
convergence of numerical methods
equiripple filters
filtering theory
iterative methods