# Accurate water quality assessment using IoNT-enabled deep learning frameworks

## TL;DR

This paper introduces a new system using nanosensors and deep learning to accurately monitor and classify water quality in real time.

## Contribution

A novel IoNT-enabled deep learning framework (WQI-CNN) for real-time water quality assessment is proposed.

## Key findings

- The WQI-CNN model outperforms existing systems in computation time, RMSE, accuracy, and MCC.
- The system achieves a high accuracy of 98.91 in real-time water quality monitoring.
- Deep GANs improve data quality by filling gaps and normalizing missing information.

## Abstract

This work proposes a novel Internet of Nano-Things (IoNT)-driven real-time system architecture of water quality (WQ) observation and classification through a Convolutional Neural Network (CNN) framework, comprising WQI-CNN. The proposed system will be organized into four stages, namely, the data acquisition, coordination, data processing, and prediction and classification of the WQ Index (WQI). State-of-the-art nanosensors, such as Luminescent TOP, Surface Enhanced Raman Spectroscopy (SERS), and graphene-based sensors are used in the sensing phase to measure important WQ parameters. The data processing step uses Deep Generative Adversarial Networks (GANs) to fill in the gap between missing information and normalize data and improve the quality of predictions. WQI-CNN model incorporates these pre-processed inputs and uses CNN to create accurate WQI classification. The system was compared with the already existing systems such as the IoT-ML, WQI-ML, GTV-STP, which showed better performance in terms of computation time, RMSE, accuracy and MCC. The WQI-CNN model can be accurately used to determine the value of a real-time WQ monitor (98.91) which is essential in the management of the proactive water under the condition of the set of the safe drinking water standards.