Hybrid optimization techniques for the design of SAW-filters

SAW filters play a key role in IF filtering in mobile communication systems. Filters based on Single Phase Unidirectional Transducers (SPUDTs) are one of the most interesting SAW structures. One aim in the design of these filters is for example to find a filter geometry with minimum length which fulfils the specifications. The high performance SAW filter designs require the solution of difficult nonlinear optimization problems. We present an optimization scheme for the search of a global optimum of these SPUDT based filters. The method consists of stochastic optimization, e.g., simulated annealing or genetic search, followed by a quasi Newton method. The combination of different optimization methods shows good results for the solution of the difficult optimization problem. The scheme leads to a good overall convergence behavior.