Centre of slice area average defuzzifier for digital fuzzy systems and its hardware implementation

Fuzzy hardware is the best choice to reach high speed inference rates in knowledge based systems. In fuzzy systems, the defuzzifier stage is the most hardware resource consumer operation and has had many different approaches with the objective of reducing the resources consumed and increasing the processing speed. From these approaches and in order to improve the defuzzifier performance, the use of integer values and the avoidance of complex operations such as multiplication and division can be noted. This paper presents a novel defuzzifier method that is designed to work with α-level based membership functions, using integer universes and avoiding multiplication operation. The result is an implementation which is simple and fully compatible with digital computers, showing the advantages of the proposed method against currently used hardware defuzzifiers such as COG and MOM.