A new approach for automated synthesis of analog and mixed-signal systems is presented. The heterogeneous genetic optimization strategy starts from a functional description and evolves a simple design solution in a strict topdown design process to a complex one that fulfills multiple objectives. Transformations of both architecture and parameters are applied. The expected improvement of the violated objectives is used as driver for the transformation selection. The topology is really created, giving the opportunity to explore new architectures.
Ewout Martens, Georges G. E. Gielen