Ambient home safety monitoring. No cameras. No wearables. Just Wi-Fi.
3rd Place — Cambridge University Press & Assessment Philippines Women in Tech Hackathon 2026
Lola Landa passed away a month before this hackathon. My aunts and titos cared for her deeply — but the reality of distance and work meant no one could watch over her 24/7, no matter how much they loved her.
That question stayed with me: Do we always just have to hope our loved ones are safe?
Project Landa is named after her. The Wi-Fi sensing nodes in the system are called Laure Nodes — after our family name.
Building something technically grounded around that real problem — the gap between love and presence — made this one of the most meaningful things I've worked on.
"We can't be there 24/7, no matter how much we love someone."
Landa is an ambient intelligence platform for passive, privacy-first home safety monitoring. It uses Wi-Fi Channel State Information (CSI) — the data already passing through your home's Wi-Fi signals — to detect falls and unusual inactivity, without any cameras or wearables involved.
Small wall-plug devices (Laure Nodes) establish invisible sensing "corridors" between a home router and receiver. When the system detects something wrong, it sends an alert to caregivers in real time.
Built for:
- Elderly family members living alone
- OFW (Overseas Filipino Workers) families monitoring loved ones from abroad
- Anyone who can't always be physically present for the people they care about
The app stays minimal and quiet under normal conditions — a deliberate "Calm Tech" design choice. When the home is secure, the interface is soft and unobtrusive (#FFF0F5 background). No noise unless there's a reason for noise.
Shows live CSI variance readings per room, room protection status, and real-time Firebase connection indicator.
When a fall is detected, the app immediately escalates to a high-contrast, full-screen alert with one-tap access to emergency contacts. No digging through menus when seconds matter.
A visual floor plan of the monitored home showing active Laure Node positions, sensing corridors, and per-room presence indicators. Caregivers can see exactly where in the home something happened.
Landa avoids false alarms by requiring both conditions to be true before triggering an alert:
- CSI Variance Spike — A sudden, catastrophic jump in CSI variance (delta > 0.65), indicating rapid mass displacement consistent with a fall.
- Sustained Stillness Window — The variance immediately drops to near-zero and stays there for 2+ seconds, confirming the person is motionless on the floor.
A single spike alone (like someone jumping or a door slamming) doesn't trigger an alert.
A dog walking through a room creates minor, rapid CSI ripples that stay below the trigger threshold (< 0.35). The algorithm distinguishes the physical signature of a 10kg animal from a human fall, keeping false alarms suppressed without requiring any configuration.
Beyond falls, the system can flag when a room shows no macro-movement over an extended period — useful for detecting incapacitation scenarios or dementia-related wandering at unusual hours.
There are no cameras, no microphones, and no wearables. Landa monitors the space, not the person. The only data captured is abstract CSI variance numbers — no images, no biometrics, no video.
A step-by-step setup flow guides users through initial configuration and Laure Node pairing. Nodes require a 10-minute calibration window to map room-specific Wi-Fi multipath fingerprints before active monitoring begins.
For development and demonstration, a Python backend simulates the ESP32 hardware, pushing realistic CSI data to Firebase. Scenario triggers (fall, pet activity, reset) are available both from the terminal and from the app UI.
| Layer | Technology |
|---|---|
| Web Dashboard | React 19 + Vite 8 |
| Mobile App | React Native 0.83 + Expo 55 (TypeScript) |
| Database | Firebase Realtime Database |
| Charts | Recharts |
| Hardware Simulator | Python + firebase-admin |
| Target Hardware | ESP32 microcontroller |
┌─────────────────────────────────────────────────────┐
│ Edge Layer │
│ ESP32 Laure Node ──or── Python Simulator │
│ (captures Wi-Fi CSI variance per room) │
└────────────────────┬────────────────────────────────┘
│ JSON payloads (1s intervals)
▼
┌─────────────────────────────────────────────────────┐
│ Firebase Realtime Database │
│ /rooms/{roomId}/csi_variance │
│ /rooms/{roomId}/status │
│ /alert_history │
│ /system_status │
│ /control/command ◄── UI trigger commands │
└──────────┬──────────────────────────┬───────────────┘
│ │
▼ ▼
┌──────────────────┐ ┌──────────────────────────┐
│ React Web App │ │ React Native Mobile App │
│ (Vite, browser) │ │ (Expo, iOS + Android) │
│ │ │ │
│ • Dashboard │ │ • Live CSI dashboard │
│ • Alert Screen │ │ • Fall alert screen │
│ • Home Map │ │ • Room status display │
│ • Admin Panel │ │ │
│ • Simulator UI │ │ │
└──────────────────┘ └──────────────────────────┘
Each room publishes a standardized payload every second:
{
"home_id": "landa_demo_001",
"rooms": {
"bedroom": { "csi_variance": 0.051, "status": "normal", "node_id": "laure_01" },
"bathroom": { "csi_variance": 0.045, "status": "normal", "node_id": "laure_02" },
"living_room": { "csi_variance": 0.088, "status": "normal", "node_id": "laure_03" }
},
"system_status": "all_secure",
"alert_history": [],
"last_sync": 1713214247
}On a fall event, the affected room's status becomes "anomaly_fall" and system_status becomes "alert_active". The frontends react immediately via Firebase onValue listeners — no polling required.
landa/
├── backend/
│ └── simulator/
│ ├── simulate.py # Python CSI data simulator
│ └── requirements.txt # firebase-admin
├── frontend/
│ ├── webapp/ # React + Vite web dashboard
│ │ └── src/
│ │ ├── App.jsx # App routing & phase management
│ │ ├── hooks/
│ │ │ └── useLandaData.js # Firebase listener hook
│ │ └── components/
│ │ ├── Dashboard.jsx
│ │ ├── AlertScreen.jsx
│ │ ├── MapScreen.jsx
│ │ ├── AdminControlPanel.jsx
│ │ ├── SimulatorPanel.jsx
│ │ ├── OnboardingScreen.jsx
│ │ ├── PairNodeScreen.jsx
│ │ └── SplashScreen.jsx
│ └── mobile/ # React Native + Expo mobile app
│ └── src/
│ ├── firebase.js
│ └── screens/
│ └── DashboardScreen.tsx
└── README.md
See GETTING-STARTED.md for full setup and run instructions.
Because CSI is sensitive to any mass displacement, the algorithm can be tuned for:
- Respiration tracking — micro-movements of the chest cavity for sleep apnea / SIDS monitoring
- Behavioral anomaly detection — location transitions at unusual hours for dementia wandering alerts
- Prolonged inactivity check-ins — flagging rooms with no macro-movement over hours for stroke/incapacitation scenarios
Landa handles the passive side of remote care — always-on environmental monitoring with zero user interaction required.
Clinivue handles the active side — cardiovascular measurement using a smartphone camera (rPPG) when the user intentionally checks their vitals.
Together, they form an "Active + Passive" remote care ecosystem: Landa detects that sleep patterns have worsened over three nights, then prompts the family to request a cardiovascular check-in through Clinivue. Moving from reactive emergency alerts to proactive preventative care.
"The camera doesn't belong on your bedroom ceiling watching you sleep — but the camera in your pocket is a powerful clinical tool when used on your own terms."
| Name | Role |
|---|---|
| Anne Reyes | Concept, architecture, development |
| Shania Dela Vega | Backend & Database Developer |
| Louella Arce Ng | Business Analyst |
| Kim Caryl Esperanza | UI/UX Designer, Researcher |
| Ms. Renilda S. Layno | Adviser |
Built for the Cambridge University Press & Assessment Philippines Women in Tech Hackathon 2026 — where we placed 3rd.
MIT License — see LICENSE for details.
Named after Lola Landa. Built with love and a bit of physics. 🤍