Project HAIP: Hybride KI intrusion prevention for industrial control systems

Challenge

In the course of the digitalization of production systems, automation systems are increasingly being networked with each other as well as with control levels and corporate networks and the Internet. These complex system landscapes result in new dangers for automation systems in terms of IT security, which are otherwise already familiar from the classic IT sector. The problem here is that existing industrial communication protocols and automation systems were often developed without sufficient protective measures such as authentication, authorization or encryption. In addition, retrofitting an existing system is complex and expensive. A potential attacker with access to the network of an automation system can carry out various attacks and cause massive damage. 

Solution approach

The overall objective of the project is to research and develop a self-learning safety solution for the protection of Industrial Control Systems. This is made up of several components:

• Learning the normal behavior of technical equipment from process data to detect anomalies
• Pattern recognition algorithm for detecting attack patterns in the communication data
• Fusion model that intelligently merges the results of the previously mentioned algorithms
• Explanation of the AI decision 
• Preparation of the results for the plant operator including recommendations for action
• Automated defenses that take effect in real time to protect the ICS
• Cloud computing aspects to exploit synergy effects 

Profile