Cyber-Physical Systems & Industrial Control Security
Think of SCADA as the "brain" that watches and controls physical machines from a central place.
A fusion of ICT (software/networks) with the physical world through sensors (input) and actuators (output).
The umbrella term for control systems used in critical infrastructure. SCADA and DCS (Distributed Control Systems) are types of ICS.
The system has 4 main hardware components that form a data pipeline from the physical world to the control center.
OLE for Process Control — software interface letting Windows communicate with industrial hardware. Like a translator between IT and OT.
| ASPECT | 💻 IT (Information Technology) | 🏭 OT (Operational Technology) |
|---|---|---|
| Focus | Data processing | Physical processes |
| Top Priority | Confidentiality | Availability & Safety |
| Downtime | Somewhat acceptable | Minimal — zero preferred |
| Updates/Patches | Frequent, automated | Slow, careful, rare |
| System Lifecycle | 3–5 years | 15–30+ years |
| Failure Impact | Data loss, financial | Physical harm, deaths |
Unlike standard IT which uses TCP/IP, ICS/SCADA uses specialized industrial protocols. Security levels vary widely.
| PROTOCOL | PORT | SECURITY | NOTES |
|---|---|---|---|
| Modbus | 502 | LOW ⚠️ | No encryption or authentication — very legacy |
| DNP3 | 20000 | MEDIUM | Supports Secure Authentication feature |
| DNP | 19999 | MEDIUM | Earlier version of DNP3 |
| OPC UA | Various | HIGH ✓ | Encrypted & authenticated — modern standard |
| IEC 61850 | Various | HIGH ✓ | Used in smart grids specifically |
| EtherNet/IP | 2222 | MEDIUM | Industrial Ethernet protocol |
| PROFINET | 34962-64 | MEDIUM | Used in factory automation |
ICS/SCADA threats are divided into two main categories:
Originate inside the premises. No malicious intent.
Intentional attacks with specific objectives.
Old system designs where security was never considered. Messages still transmitted in cleartext (no encryption).
No regular patching scheme. Some firmware is never updated. Downtime for maintenance is difficult to schedule.
Components (sensors, RTUs, PLCs) are geographically spread out — very hard to physically secure.
Managed by different departments. No coordination = security gaps. OT dept manages PLCs; IT dept manages networks. Nobody owns the overlap.
Cannot run penetration tests on production SCADA systems — the risk of disrupting real operations is too high.
Critical devices (historian servers, switches, modems, field instruments) need redundant backups to survive failures.
System should gracefully degrade:
Normal (full auto) → Emergency (partial auto) → Full
Manual (no auto)
Detect failures early — exhausted memory, bandwidth, processor usage — before they become attacks.
Modern SCADA systems are integrating AI/ML to shift from reactive to predictive & proactive operations.
| FEATURE | ⚙️ Traditional SCADA | 🤖 AI-Integrated SCADA |
|---|---|---|
| Monitoring | Reactive | Predictive & Proactive |
| Maintenance | Scheduled or manual | Predictive maintenance |
| Decision-Making | Human-driven | AI-assisted or automated |
| Anomaly Detection | Rule-based signatures | Machine learning-based |
| Data Analysis | Historical reports | Real-time intelligent insights |
| Security | Signature-based detection | AI-driven threat detection |