CryptoPulse MQTT

CryptoPulse MQTT multi-layer security framework

A multi-layer security framework for MQTT-based IoT networks combining encrypted communication, RF jamming detection, and AI-based threat analysis.

Overview

CryptoPulse MQTT (CPM) is a Taif University graduation project created to strengthen the security of MQTT-based IoT environments.

It combines secure MQTT communication, RF monitoring, AI-assisted anomaly detection, honeypot defense, and centralized security monitoring.

The project earned 1st place in the Male Students Section, 5th place overall in the Computer Engineering Department, and 3rd place in the Cultural Scientific Olympiad’s Basic & Engineering Sciences Track.

Problem

MQTT-based IoT devices face threats across insecure communication, RF jamming, signal anomalies, unauthorized access, and malicious activity.

Protecting only the network or application layer leaves physical-layer attacks uncovered—an area traditional MQTT security mechanisms may not detect.

Solution

The team built a multi-layer framework combining TLS and AES, broker authentication, RF monitoring, honeypot-based attacker detection, AI-assisted analysis, and a centralized dashboard.

My contribution focused on the AI and ML layer: I developed an ANN for RF jamming and signal-anomaly detection that achieved 91% accuracy.

I also built the end-to-end ML pipeline from data collection and cleaning through training, evaluation, and dashboard reporting and deployed the trained model to Arduino R4 edge hardware with bare-metal techniques.

Technologies

  • TensorFlow
  • Python
  • Pandas
  • NumPy
  • Scikit-learn
  • Arduino R4
  • MQTT
PosterCryptoPulse MQTT project poster showing the multi-layer security framework
Multi-layer IoT security architecture
DashboardCryptoPulse MQTT centralized security monitoring dashboard
Threat analysis and monitoring dashboard
AnalysisCryptoPulse MQTT ANN confusion matrix
ANN confusion matrix for RF jamming and signal-anomaly detection.
AnalysisCryptoPulse MQTT ANN training and validation learning curve
Training and validation learning curve for the ANN model.