Design and Implementation of a Fast General Purpose Fuzzy Processor

Design and Implementation of a Fast General Purpose Fuzzy Processor

Mohamed Ibrahim Mahmoud (Menoufia University, Egypt), Sayed Mohamed El-Araby (NRC, Egyptian Atomic Energy Authority (EAEA), Egypt), Safey Ahmed Shehata (NRC, Egyptian Atomic Energy Authority (EAEA), Egypt), Refaat Mohamed Fikry AbouZaid (NRC, Egyptian Atomic Energy Authority (EAEA), Egypt) and Fathi Abd El-Samie (Menoufia University, Egypt)
Copyright: © 2015 |Pages: 16
DOI: 10.4018/978-1-4666-7456-1.ch025
OnDemand PDF Download:
List Price: $37.50


In this paper, a Fast Fuzzy processor (FP) is proposed. This processor, which is implemented using FPGA, has four inputs and one output with 8-bits width for each. The proposed processor is synthesized, functionally verified and implemented using Xilinx Integrated Software Environment (ISE) and is tested using Xilinx Spartan 3E starter kit. A PC Graphical User Interface (GUI) is programmed using C# programming language to select and download the parameters of the processor through the serial port communication. The proposed processor is experimentally tested through water sprinkler system example. The experimental results approve the excellent performance of the proposed processor.
Chapter Preview

1. Introduction

Fuzzy inference is becoming an attractive approach to solve control and decision-making problems. This is mainly attributed to its inherent ability to manage the intuitive and ambiguous behavioral rules given by a human operator to describe a complex system. The application of fuzzy technologies to real-time control problems implies that hardware realizations can be adapted to the fuzzy paradigm (Azar, 2010). Many microelectronics implementations of fuzzy controllers have been proposed, recently. However, if fuzzy controllers are to be massively adopted in consumer products, they must fulfill some additional characteristics. First, they must be flexible, that is, suitable for adapting their functionality to different applications. This implies the capability to program the knowledge base and select different inference mechanisms. On the other hand, considering fuzzy controllers as integrated circuits, they must be efficient in terms of silicon area and operational speed (Jimenez et al., 1998).

In this paper, one of the architectures of the general-purpose fuzzy processors is designed with high speed and degree of the parallelism and programmed by VHDL code and implemented on Spartan 3E starter kit. The characteristics of the proposed fuzzy processor are shown in Table 1. It has up to 4 inputs and one output with 8-bit width for each and a number of rules depending on the memory size. The parameters of the fuzzy processor such as the number of inputs, the number of rules, and the membership values for each input, are specified by a PC GUI, which is designed for simulating the fuzzy controller depending on the application and downloading the parameters to the fuzzy processor via the PC serial port communications.

Table 1.
The proposed fuzzy processor characteristics
Fuzzy Inference SystemMamdani FIS
Input Resolution8-bit
Output Resolution8-bit
Antecedent Mf's8 Trapezoidal per fuzzy set
Antecedent MF Degree of Truth Resolution8-bit
Consequent MF's8 Trapezoidal per fuzzy set
Consequent MF Degree of Truth Resolution8-bit
Aggregation MethodMAX
Implication MethodMIN
MF Overlapping Degree2
Defuzzification MethodCenter of Gravity (COG)

Complete Chapter List

Search this Book: