A multi-layered IoT platform combining embedded sensor probes, AWS cloud intelligence, and a dual-connectivity mobile app to protect mountain communities — before disaster strikes.
Existing landslide monitoring systems fail communities at the most critical moments. We identified four core gaps.
Current systems centralize data for authorities only. Residents in high-risk zones receive warnings too late, if at all. Our system pushes alerts directly to household mobile devices in seconds.
Disaster events destroy the very internet infrastructure needed for alerts. Our dual-connection architecture ensures the app continues working via local Wi-Fi even when cellular networks are down.
Valuable soil and rainfall data collected by monitoring hardware is never exposed to researchers. Our open Research API democratizes access to anonymized environmental datasets for geological studies.
Most systems use a simple on/off alarm. Our 3-level warning logic (Normal → Warning → Dangerous) allows communities to take graduated precautions rather than reacting only when it is too late.
A vertically integrated system — from physical sensors in the ground to the cloud and into residents' pockets.
From sensor probe to resident screen — every technology chosen for a specific reason.
Every feature engineered to function when conventional systems fail.
Two sampling modes: Normal Mode polls sensors every 15 minutes to conserve battery; Burst Mode escalates to 10-second intervals automatically when vibration or tilt thresholds are crossed — capturing events in full resolution.
Aggregated sensor readings are transferred to the cloud every day via AWS IoT Core over a secured MQTT/TLS channel, ensuring a continuous and tamper-resistant time-stamped record.
All telemetry is persisted in PostgreSQL + TimescaleDB — purpose-built for high-frequency sensor data. Enables fast range queries and long-term trend analysis across months of recordings.
Anonymized environmental datasets are exposed to third-party universities and research institutes via a dedicated read-only REST API. Rate-limited and API key-gated to protect core system performance.
The EC2-hosted FastAPI backend validates incoming payloads, runs multi-parameter risk analysis, and generates predictive landslide probability scores to drive intelligent alert decisions.
Alert rules are fully configurable per probe zone. Authorities receive notifications across three warning levels — Normal, Warning, and Dangerous — with escalating notification channels.
A Flutter mobile app with a registration system lets households monitor their local slope in real time. Long-term historical graphs let residents track gradual ground movement trends over weeks and months.
A multidisciplinary third-year project team spanning embedded systems, cloud infrastructure, and mobile development.
Estimated component costs for a single SlideSense probe unit in Sri Lankan Rupees (LKR).
| Component | Price (LKR) |
|---|---|
| 20W 17.87VDC Solar Panel | 3,350.00 |
| 1200mAh 3.7V 383562 Lithium Li-Polymer Rechargeable Battery (Li-Po) | 1,185.00 |
| CN3791 12V MPPT Solar Charge Controller | 1,450.00 |
| 3.3V Regulator | 120.00 |
| Capacitive Soil Moisture Sensor (LKR 290 × 3) | 1,160.00 |
| Tipping Bucket Rain Gauge | 10,000.00 |
| High Sensitivity Microphone Sensor Module | 200.00 |
| SIM-900A | 1,450.00 |
| GPS-NEO-M8N | 2,350.00 |
| Grand Total | 21,265.00 |
