It's different: Insights into home energy consumption in India

Nipun Batra, Manoj Gulati, Amarjeet Singh, Mani Srivastava

Special thanks: Milan Jain, Shailja Thakur

NB: You can view a prettier rendering of the same page here.

Deployment Overview

The figure above shows the overall deployment across a 3 storey home in Delhi, India. We deployed a total of 33 sensors measuring the following different parameters:

• Electricity
  • Household level (using Schnieder Electric EM6400 smart meter)
  • MCB level (using custom microcontroller based CT platform)
  • Appliance level (using jPlug, developed by Radio Studio, India)
• Water
  • At inlet from utility and outlet from water tank (using Pulse based water meters)
• Ambient parameters
  • Motion (using Express Controls HSM100)
  • Light, temperature (using Express Controls HSM100 and FunF journal on Android phones)
• Connectivity and Network (using Android, soft sensor streams)
• Weather (using 3 weather monitoring stations providing data via REST APIs)

Repository Structure

This repository is divided as follows:

• Configuration
  • Database schemas: Schemas for MySQL databases used in the deployment
  • Misc: Miscellaneous mappings between sensor id and locaiton in home
  • Configuration settings: of different sensors. eg. sensitivity of HSM100
  • Inventory: Detailed inventory mentioning all the resources used
• Dataset
  • Annotated dataset: Contains fully labeled dataset from our deployment.
  • Metadata: Contains pertinent metadata for different appliances in the home and different water fixtures
• How To: Misc. how to guides which are not specific to the deployment, but used in our case
• Software:
  • Common: Software common to SBCs and Server
  • Sensors_SBC: Software used by different sensors/SBCs to get data from sensors
  • Server: Software used for server management
  • Startup scripts: Various startup scripts used for uninterrupted data collection (even after power outage)
  • Installation: How to prepare the various computational systems for deployment
  • Stats: Scripts and their outputs used for the plots used in the paper
• Deployment notes

One can also find the various issues and how we resolved them.

1. Open Issues
2. Closed Issues

Dataset

1. The detailed metadata dataset can be found here.
2. The UTC timestamped annotated dataset can be found here. The event sequence for the same and more information about the same can be found on the same page. This can be visualized here.

The figure below shows the utlity of multi-modal time synchronized annotations, from one of the days during our deployment.