Cost-efficient implementation of adaptive finite state machines

This paper concerns the implementation of adaptive controllers, in particular Finite State Machines (FSMs) that change their behavior at run-time. The main emphasis is to develop a technique for creating very dense-coded FSMs in order to reduce SoC programming cost. We propose an approach that combines memory-based function generators with a programmable logic to achieve proper results. For experimental tests of our FSM core we utilized a Xilinx Virtex FPGA.