Technical depth / education

ECG RR-interval alert (edge to cloud)

An EE445 group project. Embedded C++ pipeline on an MSP432P401R doing real-time single-lead ECG analysis with Pan-Tompkins QRS detection, pushing MQTT alerts into ThingsBoard Cloud when the RR interval drops below 0.6 seconds.

Employer / client: Dublin City University
Duration: EE445 group project, April 2024
Project type: Embedded systems project

GitHub ecg-ventricular-detection

Architecture

How the edge-to-cloud pipeline works

Edge-to-cloud signal pipeline: a single-lead AD8232 sensor feeds the MSP432 microcontroller, which runs Pan-Tompkins QRS detection on-device. When the RR interval drops below 0.6 seconds it trips an alert published over MQTT to ThingsBoard Cloud. Edge detection stays separate from cloud ingestion. — How the single-lead ECG signal is detected on the MSP432 at the edge, then turned into an MQTT alert in ThingsBoard Cloud when the RR interval crosses 0.6 seconds.
My edge-to-cloud architecture

Process flow

How I work the steps

01
before Acquire signal
AD8232 single-lead ECG into the MCU.
Sensor
02
control Edge detection
Pan-Tompkins QRS detection runs on the MSP432.
My signal-processing piece
03
risk Threshold check
RR interval under 0.6 s trips a high-rate alert.
Detection logic
04
after Cloud alert
MQTT alert lands in ThingsBoard for review.
ThingsBoard Cloud

How I built it

Balanced MCU memory and timing constraints against the real-time ECG sample rate.
Kept the edge detection logic separate from the cloud ingestion so failure modes were easy to reason about in demo and review.
Threads straight into my interest in governed operational data pipelines: signal processing on the edge, alerting in the cloud, audit trail in between.

Measured results

What I measured

~85% evidence

RR-alert detection accuracy (project setup)

Implemented Pan-Tompkins QRS detection streaming on the MSP432P401R against a single-lead AD8232 ECG sensor.

Findings

Implemented Pan-Tompkins QRS detection streaming on the MSP432P401R against a single-lead AD8232 ECG sensor.
Hit about 85% accuracy on RR-threshold alert detection in the project test setup.
Shipped MQTT-based alerting into ThingsBoard Cloud, triggered when the RR interval crossed the 0.6 s threshold.