Automated architectural optimization of digital FIR filters

Using a newly developed fine-grained architectural synthesizer for digital finite impulse response (FIR) filters, the authors present a near-optimal strategy for coefficient manipulation that runs linearly with filter length. The synthesizer operates at a high level of abstraction, balancing speed, area and power goals to generate a circuit optimized for specified parameters. Experimental results show that it is highly competitive with the best efforts of human designers and other automated procedures.