An symbolic decomposition of functions with multi-valued inputs and outputs for FPGA-based implementation

In this paper a method for symbolic decomposition of functions with multi-valued inputs and outputs is presented. Decomposition is performed simultaneously with an encoding of symbolic values. In this way an impact of input/output encoding on decomposition efficiency is taken into consideration during optimization. The input/output encoding is built in the balanced decomposition strategy based on parallel and serial functional decompositions. The experimental results show that the presented method significantly reduces the cost of a FPGA implementations for most evaluated benchmarks.