CIC IIoT Dataset 2025 - DataSense Real-Time Sensor Benchmark

Dataset Overview

The CIC IIoT Dataset 2025 (DataSense) represents the latest advancement from the Canadian Institute for Cybersecurity at University of New Brunswick. Released in December 2024, this dataset addresses critical gaps in existing IIoT security research by providing synchronized sensor and network traffic data from a realistic industrial testbed.

Testbed Architecture - 5 Layers

The sophisticated experimental environment comprises 40 interconnected devices organized across five primary layers:

1. IoT/IIoT Layer

Over 15 types of industrial sensors including custom-built Arduino-based sensors and real industrial-grade devices monitoring temperature, pressure, vibration, flow rate, level detection, proximity, and environmental parameters.

2. Network Infrastructure

Enterprise-grade switches, routers, firewalls, and industrial protocols (Modbus TCP, OPC UA, MQTT) replicating real factory networks.

3. Edge Layer

Edge computing devices performing local data processing, protocol translation, and real-time analytics before cloud transmission.

4. Cloud Layer

Cloud platforms for data storage, advanced analytics, and centralized monitoring dashboards.

5. Attacker Layer

Dedicated systems simulating sophisticated cyber-physical attacks targeting industrial control systems.

Novel Feature Selection Method

The dataset introduces a multi-objective feature selection approach that:

• Enhances anomaly detection accuracy beyond traditional methods
• Minimizes computational resource usage for edge deployment
• Balances detection performance with real-time processing requirements
• Reduces false positive rates in industrial monitoring systems

Data Synchronization

A key innovation is the precise synchronization between sensor readings and network traffic, enabling correlation analysis between physical measurements and cyber events. This allows researchers to study cyber-physical attack patterns where network intrusions manifest as anomalous sensor behaviors.

Attack Scenarios

The dataset captures realistic industrial attack vectors including:

• Sensor spoofing and data manipulation
• PLC (Programmable Logic Controller) compromise
• SCADA protocol exploitation
• Network intrusions affecting operational technology
• Denial of service targeting industrial processes
• Man-in-the-middle attacks on industrial protocols

Research Applications

• Developing IIoT-specific intrusion detection systems
• Cyber-physical anomaly detection research
• Feature engineering for resource-constrained edge devices
• Testing resilience mechanisms in industrial environments
• Evaluating federated learning for distributed IIoT security
• Benchmarking AI models for operational technology protection

Academic Recognition

Published in Electronics journal (2025) by Firouzi et al. with DOI reference. The dataset brings academic advances closer to practical IIoT security solutions deployable in real industrial settings.