High-level view of all system components and how they interconnect across the eCozy ecosystem.
Edge devices (thermostats, climate sensors) communicate over Thread via Matter. The Central Unit is the gateway between edge devices and the cloud. Currently all control flows through the Cloud Platform (AWS) — the Mobile App sends requests to the cloud, which forwards them to the Central Unit. Direct Wi-Fi access from the app (Central Unit as hotspot) is planned.

| Component | Description | Key Technologies | Protocol / Transport |
|---|---|---|---|
| Thermostat | Measures room temperature and humidity (SHT40), calculates PID, actuates radiator valve. On-device fire alarm detection (Edge AI). LCD display. | Nordic nRF5340, Zephyr RTOS, TFLite Micro | Matter-over-Thread |
| Climate Sensor | High-precision temperature and humidity readings (SHT40) alongside the thermostat. Ultra-low-power, coin-cell powered. | Nordic nRF52840, Zephyr RTOS | Matter-over-Thread |
| Central Unit | Gateway between Thread network and the cloud. Buffers telemetry, distributes OTA to edge devices. Acts as Thread Border Router. Direct Wi-Fi control from the app (hotspot mode) is planned. | RPi CM4 + nRF52840, Yocto Kirkstone | Matter-over-Thread ↔ HTTPS / REST |
| Cloud Platform | NestJS REST API, PostgreSQL database. Handles user commands, telemetry, device management, and OTA rollouts. | NestJS, PostgreSQL, hawkBit, AWS | HTTPS / REST (TLS 1.3) |
| Mobile App | Primary user interface — device onboarding, room control, schedules, energy analytics, push notifications. | Flutter (Dart), BLoC, Clean Architecture | HTTPS / REST + FCM / APNs |
For detailed documentation on each part of the system, see the pages below.
| Page | Key Technologies | Description | |
|---|---|---|---|
| 🔩 | Hardware | Nordic nRF5340, nRF52840, RPi CM4 | Device inventory, chipsets, connectivity standards, power & energy design, deployment model. |
| ⚙️ | Firmware / Embedded | Zephyr RTOS, Yocto Kirkstone, SWUpdate + hawkBit | OS/RTOS per device, modular firmware architecture, OTA update mechanism, security architecture. |
| 🧠 | Edge AI | TensorFlow Lite Micro, INT8 CNN | On-device audio classification (fire alarm), inference pipeline, compute constraints, roadmap. |
| ☁️ | Cloud | AWS, NestJS, PostgreSQL, hawkBit | Backend API, data storage, device management, OTA, voice assistants, API security. |
| 📱 | Application | Flutter/Dart, BLoC, Matter | Mobile apps, dashboards, REST API, smart home integrations (HomeKit, Google Home, Alexa). |
| 🔄 | Communication Flow | Matter-over-Thread, HTTPS / REST | End-to-end data paths, protocol details, sequence of operations, offline behavior. |
| 📊 | Data Flow & Decision Logic | — | How data moves through the system and the logic behind automated climate decisions. |
| 🔌 | Offline vs. Cloud | — | What the system can do locally and what requires cloud services. |
| 🏗️ | Backend Infrastructure | — | Server architecture, APIs, and cloud services powering the platform. |