From Dam to Tap in Mumbai - Real-Time Water Quality Monitoring using IOT and Machine Learning

Aug 5, 2026

Upload by Sage

This research explores real-time water quality monitoring across the supply chain from dams to households using IoT sensors and machine learning. Water samples are collected at the Bhandup Complex for purification and then distributed to 26 service reservoirs in Mumbai. Data from reservoirs and purified water trains and tests models including SVM, Random Forest, and LightGBM, with LightGBM achieving 97.5% accuracy on purified water and 75.5% on reservoir water. The AquaSage approach supports early detection of contamination, improved public health protection, and sustainable water management, while noting limitations in dataset coverage and sensor scalability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Literature Review

AquaSage and proposed solution

20(1): 445-454, 2025 [www.thebioscan.com](www.thebioscan.com) From Dam to Tap in Mumbai: Real-Time Water Quality Monitoring using IOT and Machine Learning 1 Pearl Dsouza, 2Namrata Joshi,3 Noelle Shaji, 4 Prof. Prachi Patil, 5 Prof. Monali Shetty 1,2,3, Student, Department of Computer Engineering, Fr. Conceicao Rodrigues College of Engineering. Mumbai 4,5, Assistant Professor, Department of Computer Engineering, Fr. Conceicao Rodrigues College of Engineering. Mumbai [1](1 crce.9538.ce@gmail.com)[ ](1 crce.9538.ce@gmail.com)[crce.9538.ce@gmail.com](1 crce.9538.ce@gmail.com), [2](2 crce.9545.ce@gmail.com)[ ](2 crce.9545.ce@gmail.com)[crce.9545.ce@gmail.com](2 crce.9545.ce@gmail.com), [3](3 crce.9577.ce@gmail.com)[ ](3 crce.9577.ce@gmail.com)[crce.9577.ce@gmail.com](3 crce.9577.ce@gmail.com), [4](4 prachip@fragnel.edu.in)[ ](4 prachip@fragnel.edu.in)[prachip@fragnel.edu.in](4 prachip@fragnel.edu.in),~~5 ~~[shettymonalin@gmail.com](shettymonalin@gmail.com) DOI: [https://doi.org/10.63001/tbs.2025.v20.i01.pp445-454](https://doi.org/10.63001/tbs.2025.v20.i01.pp445-454) KEYWORDS water quality monitoring, public health, and environmental sustainability, machine learning algorithms, IOT Received on: 04-01-2025 Accepted on: 03-02-2025 Published on: 06-03-2025 ABSTRACT This research explores water supply from dams to households at real-time monitoring of the quality of the water with IoT sensors and subsequent machine learning. It ensures that water supplied from dams to reservoirs and further to household service remains pure even after its supply. The methodology includes the collection of water at the Bhandup Complex for purification and its subsequent distribution to 26 service reservoirs throughout Mumbai. For training and testing different machine learning algorithms, namely SVM, Random Forest, and LightGBM, data from these reservoirs and purified water in the Bhandup Complex are used. LightGBM achieved high accuracy on the other algorithms and achieved an accuracy of 97.5% on purified water and 75.5% on the corresponding reservoir water. Real-time monitoring of water quality by IoT sensors in houses can give information on pH, turbidity, TDS, and many other parameters. It can help trace out specific areas of contamination in the water distribution network further. Major Findings for the present study comprised the reliability of AquaSage to predict water quality and the presence of potential points of pollution, thereby ensuring that the water supply from the treatment plant would be safe for consumption. This technology brings to light the major life improvements for the citizens of public health through prompt intervention and the provision of timely water quality. Thanks to their features, one of which is the possibility of an early stage of knowing that contamination exists, they are also cutting the risk of waterborne diseases by limiting the growth of bacteria and such, so they area great way to help sustainable water management also. On the other hand, the research shows some limitations. The data for dams and reservoirs, based on BMC's provided ranges, covers only five years and is static, not real-time. Efficiency of hardware sensors may decrease over time, and the newest scaling is possible only at households, but not in society or at building level. With this in mind, the major benefits of the study come in terms of creating an IoT infused machine learning algorithm for monitoring the projects in real-time and therefore using a user-friendly system that empowers citizens to ensure water safety. Practical implications involve eliminating the threats of waterborne illnesses in the city through giving the city dwellers an honest mechanism to avail clean water supply. The real-time monitoring features of AquaSage also prove useful for precautionary measures against contamination and sustainable water management. Later studies should focus on refining precision in forecasting, additional

FAQs

What is the main goal of the AquaSage system in this study?

The study aims to monitor water quality in real time from dams to households, helping ensure supplied water remains pure during transit and improves public health protection.

Which machine learning algorithms are evaluated and what results are reported?

SVM, Random Forest, and LightGBM are evaluated. LightGBM performs best, reaching 97.5% accuracy on purified water and 75.5% on reservoir water.

How does the system help identify contamination risks?

Real-time IoT-based monitoring provides metrics such as pH, turbidity, and TDS, which can help trace contamination points across the distribution network.

Aug 5, 2026

Upload by Sage