IoT Projects in Bangalore

An ESP32-based sensor node measures soil moisture, ambient temperature, humidity, and light intensity in a greenhouse environment. Readings are published every 30 seconds via MQTT to an AWS IoT Core broker. A Lambda function processes incoming data and triggers a relay-controlled solenoid valve when soil moisture drops below a configurable threshold. A React dashboard displays real-time sensor readings, irrigation event history, and a 7-day trend chart. The system reduces water consumption by an estimated 35% compared to fixed-schedule irrigation in a 30-day field trial.

An accelerometer (MPU-6050) and temperature sensor (DS18B20) are mounted on a motor to capture vibration signatures and thermal data at 100 Hz. An ESP32 performs edge FFT computation to extract dominant frequency components before transmitting feature vectors to an MQTT broker. A Python anomaly detection model running on a Raspberry Pi 4 classifies the motor state as normal, imbalance, or bearing fault with 93% accuracy. Alerts are sent via Telegram bot when a fault is detected. The project includes a Grafana dashboard showing real-time FFT spectra and fault history.

Non-intrusive load monitoring is implemented using a current transformer (SCT-013) and voltage sensor (ZMPT101B) connected to an ESP32 to measure real and reactive power consumption of individual appliances. Power readings are published to an MQTT broker and stored in InfluxDB. A Grafana dashboard displays hourly, daily, and monthly energy consumption per appliance with cost estimates based on BESCOM tariff rates. A rule engine triggers WhatsApp alerts when daily consumption exceeds a user-defined budget. The system achieves ±3% accuracy compared to a calibrated energy meter.

Four sensor nodes, each comprising an ESP32 with MQ-135 (CO2/VOC), DHT22 (temperature/humidity), and PMS5003 (PM2.5/PM10) sensors, are deployed across a college building. Data is transmitted via Wi-Fi to a central Raspberry Pi running an MQTT broker and Node-RED flow processor. A web dashboard built with Grafana displays real-time AQI for each room with colour-coded health risk levels. Automated ventilation recommendations are generated when CO2 exceeds 1,000 ppm. The system logged 30 days of data showing post-lunch CO2 spikes in classrooms with poor ventilation.

A compact tracking device built with an ESP32, NEO-6M GPS module, and SIM800L GSM module reports the location of college laptops and projectors every 5 minutes via HTTP POST to a Node.js backend. Locations are stored in MongoDB and displayed on a Leaflet.js map with movement history trails. Geofencing alerts are triggered via SMS when an asset leaves the campus boundary polygon. The device runs on a 2,000 mAh LiPo battery with a 72-hour life in tracking mode. The project includes a tamper-detection accelerometer that triggers an immediate alert if the device is moved abruptly.

Ultrasonic sensors (HC-SR04) mounted inside dustbins measure fill levels and transmit data via LoRa (SX1278) to a central gateway ESP32 connected to the internet. A Node.js backend aggregates fill-level data from 10 simulated bins and runs a route optimisation algorithm to generate the most efficient collection route for the garbage truck. A React dashboard displays a campus map with colour-coded bin fill levels and the optimised collection route. The system reduces unnecessary collection trips by 40% in simulation compared to a fixed daily schedule.

A wearable device built on an Arduino Nano 33 BLE Sense measures heart rate (MAX30102), SpO2, and body temperature (MLX90614) and detects falls using the onboard IMU. Vital signs are transmitted via Bluetooth Low Energy to a companion Android app that forwards data to a Firebase Realtime Database. A caregiver web dashboard displays live vitals with configurable alert thresholds. The fall detection algorithm achieves 96% sensitivity and 91% specificity on a dataset of 200 simulated fall and non-fall events. SMS alerts are sent to emergency contacts within 10 seconds of a detected fall.

IR sensors (IR-FC-51) detect vehicle presence in 20 parking slots. An Arduino Mega aggregates sensor states and transmits slot occupancy data via ESP8266 Wi-Fi module to a Firebase Realtime Database every 5 seconds. A React web app and Android app display a colour-coded parking map showing available (green) and occupied (red) slots in real time. An entry gate servo motor opens automatically when a free slot is available, detected by a vehicle presence sensor at the entrance. The system reduces average parking search time from 4.2 minutes to 1.1 minutes in a 2-week campus trial.

An ultrasonic water level sensor and tipping-bucket rain gauge connected to an ESP32 monitor river level and rainfall intensity at a simulated flood-prone location. Data is published to an AWS IoT Core topic every minute. A Lambda function evaluates three alert levels — watch, warning, and emergency — based on configurable thresholds and sends SMS alerts via AWS SNS to registered subscribers. A public-facing React dashboard displays current water level, rainfall rate, and alert status with a 24-hour historical chart. The system is validated against historical flood event data from the Karnataka State Disaster Management Authority.