About the Client

The client is a Alabama, USA based start-up that is developing technology to combine artificial intelligence with smart devices and the internet of things to monitor and control residential and commercial water usage. The technology will help to boost water efficiency, reduce insurance claims due to water damage, and reduce demands on local water supply and wastewater treatment facilities.

About the Project

The client needed a complete platform which could solve the following use cases:

  • Connect an IoT enabled flow meter to the water pipes
  • View the water usage data over Cloud, consumed through a mobile app
  • Detect anomalous water flow and shut the flow if needed
  • Automatically shut off the flow when the user does not respond to alerts
  • Learn flow patterns using machine learning and predictive analytics.
  • Regularly capture the flow rate and transmit relevant information to the analytics engine 
  • Control water flow from the web and mobile application

Challenges

Project Requirements include;

  • To develop an IoT enabled smart water monitoring device that can report water flow and push user notifications based on a machine learning model of AI-based dataramp architecture.
  • To gather a large amount of raw data from various use cases that are required for making decisions on flow rates by the rule engines running on the Dataramp platform.
  • To do Real-time prediction on high-velocity data generated from a huge array of IoT devices.
  • Visualise this data and integrate it with the apps and platform.

ArkLabs needed a complete platform with the following functionalities;

  • Connect an IoT enabled flow meter to the water pipes
  • View the water usage data over the cloud, consumed through a mobile app
  • Detect anomalous water flow and shut the flow if needed 
  • Automatically shut off the flow when the user does not respond to alerts
  • Learn flow patterns using machine learning and predictive analytics using a cloud-based AI platform
  • Regularly capture the flow rate and transmit relevant information to the analytics engine 
  • Control water flow from the web and mobile application

Implementation

The project is to develop an IoT enabled Smart Water Monitoring Device that can detect the water flow. It leverages Artificial Intelligence and will notify if there is a significant abnormality and can shut off the water supply. The smart water monitoring device is based on a machine learning model of AI-based dataramp architecture. The real-time data is collected from the flow meter which has components such as water flow sensors and these data will be sent to the particle cloud when the device detects the water flow, which has the corresponding device id along with the water usage and time. These data will be fetched by the data interface module of the dataramp from the particle cloud via rest APIs and get pushed into the Kafka topic.

The data pipeline component of the dataramp will receive device data in a distributed manner and will be buffered in Kafka. This means the data which comes in the dataramp platform will be divided into a numbered stream of ordered messages and stores in different topics like water usage, device id, user id, etc.

The batch processing and stream processing capabilities of the dataramp architecture will be leveraged to employ a predictive algorithm to predict various water flow scenarios. Further processing the dataramp will introduce data into the spark streaming which contains the ML module, here receivers of the spark streaming will receive and chop up the data streams into batches and send to the  ML module, where ML module contains ML model (which will be previously persisted to the data store) and score Model and it will push out the result to an external data stores or to the dashboard using REST endpoints. Datastore(elasticsearch) will consume the data in 2 minutes from a Kafka topic and push it to the web browser using socket connection for displaying the water flow.

The Solution

Built a complete supply chain solution with the following components:

A complete IOT based platform to save water was custom built with the following components

  • Web application
  • iOS App
  • Android App

Following are the highlights of the solution

  • IOT enabled water flow meter was created using an ultrasonic water flow meter to integrate with Particleboard
    • Native mobile apps were built using iOS and Android framework
    • A multi-tier architecture was developed using NodeJS

Architecture Diagram

Screens

Technology Stack Used

  • Front-End – AngularJS, HTML5, CSS, BootStrap
  • Back-End – NodeJS, StrongLoop
  • Database – MongoDB
  • Hosting – Amazon Web Services
  • Android – Java, XML
  • iOS – Objective-C
  • Analytics – Apache Kafka, Apache Spark, Elasticsearch

Looking for a similar App ?