IoT Fall Detection Monitoring

Three-layer distributed system for automatic fall detection in elderly people. At the edge, an ESP32 with an MPU6050 accelerometer detects the impact and publishes telemetry over MQTT; the middleware (ThingsBoard CE + PostgreSQL on Docker) stores the time series and triggers the alarm; a Next.js dashboard displays falls and performance in real time. To validate scale without physical hardware, I built a Python load generator (asyncio/aiomqtt) that simulates thousands of sensors, with a distributed mode across containers. With the middleware on AWS and load published from separate machines, the system sustained 1,000 devices at 5,069 msg/s with a p99 latency of 147 ms and zero errors. Academic team project (Distributed Systems — Computer Engineering, UFMA).