Hardware implementation and comparison of new defuzzification techniques in fuzzy processors

This paper deals with hardware implementation aspects of the defuzzification block in fuzzy controllers and processors. Three new defuzzification methods are introduced which are suitable for low cost hardware implementation. A complete set of common existing defuzzification methods are reviewed to be compared with these new methods from different hardware implementation aspects. Two different hardware models with different structures are developed for each method. The first model realizes the full combinational or fastest possible hardware implementation, and the second model demonstrates the full sequential or the smallest possible hardware implementation of each method. All models are synthesized on 0.18 micron CMOS technology cells to analyze and compare the area, delay and power consumption of different realizations of defuzzification methods. Some area-power-delay-accuracy analysis diagrams are then introduced according to the synthesis results to guide the designers through choosing the defuzzification method and also the implementation structure which best suites their application from implementation cost, speed, power consumption and output accuracy points of view