Low complexity artificial neural network unit for sugar content detection in microwave sensor system

Abstract

This research presents a low complexity artificial neural network (ANN) unit that used for implementation of microwave sensor system in order to detect sugar content in tested solution. Generally, the hardware implementation of ANN suffers from the number of multiplication that used in network and the difficulty of implementing sigmoid activation function. In this paper, distributed arithmetic (DA) will be used to represent the multiplication and make our ANN unit multiplier less computation unit. An efficient digital implementation of sigmoid activation function by using piecewise linear approximation will be used. Therefore, low complexity ANN unit can be achieved and it will be used in a microwave sensor system for sugar content detection.