Optimum Design of Cascaded Digital Filters in Wideband Wireless Transmitters using Genetic Algorithms

In digitally intensive direct conversion transmitters, the baseband data is up-sampled to the RF rate. As the bandwidth of the baseband data increases, carefully designed cascaded digital filters are required in order to attenuate the wide replicas generated during the digital up-sampling process. Though design methodologies for single stage digital filters are very well established, near-optimum design of such multistage filters typically requires selection of several parameters by trial-and-error. This approach is time-consuming and does not assure the optimum solution (i.e. lowest area/power) under various performance constraints. In this paper, a genetic algorithm (GA) based generic automated search methodology for design of such cascaded filters is proposed. The proposed technique is demonstrated for digital WiMAX and WCDMA transmitters, for which near-optimal solutions appear to have been achieved in a relatively short time compared to the traditional manual design techniques.