Securing IoT Devices
Machine learning applied to connected-device security: an anonymous IoT dataset used to train a neural network and a convolutional model to spot anomalous device behaviour, with the datasets and model code alongside the write-up.
Securing the Internet of Things
Executive Summary
As a cybersecurity specialist, understanding how IoT data is collected and processed is critical for protecting individuals and nations from data theft and manipulation. This project uses an anonymous IoT dataset to build machine learning models — including a Neural Network and Convolutional Neural Network — that predict user actions from IoT device data. The work demonstrates practical application of ML to IoT security challenges including threat detection and resilience against adversarial attacks.
Key Highlights
Machine Learning Models
Neural Network and CNN models built to predict user behaviour from IoT device data.
IoT Threat Analysis
Evaluation of how adversaries exploit IoT data including Trojan apps and device manipulation.
Dataset Analysis
GPS and IoT datasets processed and analysed to extract actionable security insights.
Critical Infrastructure
Assessment of IoT risks to critical infrastructure and recommendations for resilience.
Academic Report (Word)
Full academic documentation including research, implementation details, testing methodology, and evaluation.
Download Word Document →IoT Presentation
Datasets, Neural Network & Machine Learning Algorithms
GPS Dataset (CSV)
Download GPS Dataset →Neural Network (Python)
Download Neural Network →Convolutional Neural Network (Python)
Download CNN ML →Technical Breakdown
- IoT data collection, processing, and anonymisation techniques
- Neural Network model design and training on IoT datasets
- Convolutional Neural Network implementation for behaviour prediction
- Threat modelling and adversarial attack analysis on IoT systems
- Critical infrastructure risk assessment and mitigation strategies