Replacing Inputs of Look-Up-Table-Based Moore Finite State Machines with Two Cores of Input Memory Functions

A new architecture and design method are proposed for field-programmable gate array (FPGA)-based Moore finite state machines (FSMs). The proposed architecture includes a core based on functional decomposition, a core based on twofold state assignment (TSA), and a block for replacing FSM inputs (RI) in the TSA-based core. TSA belongs to the class of structural decomposition methods. The conditions under which the proposed method leads to FSM circuits with better spatial characteristics than their counterparts without RI are identified. This improvement is accompanied by a degradation of temporal characteristics. The method targets Moore FSM circuits implemented with look-up table (LUT) elements. The TSA-based core is based on classes of compatible states. To reduce the number of these classes, we use the replacement of FSM inputs by additional variables. The efficiency of the proposed method is evaluated using experiments based on a known library of benchmark FSMs. The experiments show that the trade-off between LUT count and propagation time depends on the FSM group. For the SD group (14 benchmarks), the LUT count is reduced by 13.28%, while the propagation time increases by 41.08%. For the FD group (6 benchmarks), the LUT count is reduced by 8.09%, while the propagation time increases by 2.88%. Over all 53 benchmark FSMs, the LUT count is reduced by 2.20%, while the propagation time increases by 18.77%.