A novel state assignment method for Extended Burst-Mode FSM design using genetic algorithm

With the evolution of microelectronics, more and more high-complexity digital systems are been designed. Once the global clock signal is a main concern for synchronous systems design, asynchronous circuits seem to be an interesting alternative, once they don't present clock skew and clock distribution problems. However, the lack of tools for automatic synthesis is a major drawback. This paper proposes a new algorithm for state assignment of Extended Burst-Mode Asynchronous Finite State Machines (XBM_AFSM), which are widely used in the design of Controllers of asynchronous digital systems. The proposal is based on genetic algorithm and introduces a novel style of state assignment. It improves the results and overcomes the previous methods found in literature once it addresses the “state minimization”, the “free of critical race coding” and “coverage” as a single problem. Furthermore, it is able to detect the conflicts in XBM specification and insert the minimum number of state variables in the XBM specification to eliminate those conflicts. A dedicated computational tool called SAGAAs implements the algorithm and was tested to a set of 36 benchmarks. When compared to 3D tool, our method achieved an average reduction of 13.33%, 35.65%, 32.15% and 39.12% in the number of state variables, number of products, number of literals and the number of switching, respectively. When compared to Minimalist tool, SAGAAs provided an average reduction of 52% in the number of inserted state variables, 11% in the number of literals, and a penalty of 8% in the number of products. Results show that the method and the dedicated computational tool SAGAAs achieved good and reliable results, showing a high potential of practical implementation in actual circuits design.