Neuro-ASIC for low cost supervision of water pollution

The design of a neural ASIC is presented that implements a low-cost system for the supervision of water quality in urban canalization or rivers. A trainable multilayer perceptron estimates the parameter COD (chemical oxygen demand) which is generally used to estimate water quality. The system detects correlations in the signals of low-cost sensors. It constitutes a significant cost reduction in the supervision of water environment pollution. The paper focuses on the electronic implementation