A field programmable Gm-C filter array (FPAA) for online adaptation to environmental changes.

This work presents an field programmable analog array (FPAA) architecture for reconfigurable implementation of continuous-time (CT) analog filters. First, a Matlab model of the FPAA is created and verified for compliance with transistor-level simulations. Using this model it is shown how the versatile architecture of the FPAA provides the basis for immediate adaption of filter structures to changing requirements. Secondly, a robust genetic algorithm is presented, which allows synthesis of analog filters on the given structure. It can be used to compensate for process variations or to overcome quantization limitations, as well as circumvention of hardware defects. The system is not only a platform for theoretical investigation of filter structures on the given chip structure but also provides a framework for evolution and instantiation of filters on actual chip hardware.