Smart Water Monitoring

Raspberry Pi + IoT sensors for real-time usage tracking and anomaly detection.

Raspberry PiPythonMQTTInfluxDBGrafana

Live Monitoring (Simulated)

Flow Rates (L/min)LIVE

2.5

Main Line

0.0

Shower

0.0

Outdoor

Today's Usage

152.3 liters

0 LTarget: 200 L/day

Edge-to-Cloud Architecture

Edge Layer

YF-S201 SensorsGPIO InterruptsPulse Counter

Gateway (Pi)

Python DaemonSQLite CacheAnomaly Detection

Cloud

MQTT BrokerInfluxDBGrafana

SensorsPi HubMQTTDashboard

Anomaly Detection

Z-Score Based Detection

7-day rolling baseline per hour. Alerts when flow > 2σ from expected.

z = (current - μ) / σ
anomaly = |z| > 2.0

Continuous flow > 30 min triggers leak alert via Pushover API.

Data Pipeline

Pulse Detection

Hall-effect sensor generates pulses (~4.5 per mL)

GPIO Interrupt

Pi captures rising edges with µs timestamps

Flow Calculation

Pulses/sec × calibration → L/min

Local Storage

SQLite buffer for offline resilience

MQTT Publish

60-second aggregates to cloud broker

Design Decisions

MQTT over HTTP

Lightweight protocol with built-in QoS, persistent sessions, and 99.8% bandwidth reduction vs raw telemetry.

Edge Analytics

Anomaly detection on Pi enables offline alerting and reduces cloud dependency.

Per-Sensor Calibration

Factory calibration varied 5-12%. Field calibration improved accuracy from ±10% to ±2%.

SQLite Buffer

30-day local history prevents data loss during outages, enables offline queries.

±2%

Sensor Accuracy

60s

Cloud Sync Interval

30min

Leak Alert Threshold

15-30%

Potential Savings