Reinforcement Learning Model Free with GLIE Monte-Carlo on Policy Update for Network Topology Discovery

Providing cybersecurity for computer networks is one of the main concerns for companies in this digital society. Unless this is done, companies can potentially lose huge amounts of data or even lose control of their own computer networks. Knowing the topology of the computer network and the information that is accessible on each of the nodes of the network is very relevant information both to establish impenetrable cyber defenses and to spread malware through the network and take control. In the proposed work, an algorithm has been designed to control the propagation of a malware through an unknown computer network in order to extract the information of its network topology. The results of this new algorithm have been tested in 3 simulations on a virtual copy of the same real computer network of an intelligent building in the city of Salamanca. The main result obtained was that the algorithm is able to discover all the nodes of the network adapting itself to the network characteristics.

Automated summary

Providing cybersecurity is crucial for companies in the digital society to prevent data loss or loss of control over computer networks. Understanding the network topology and accessible information on each node is essential for establishing reliable defenses and spreading malware. This research presents a new algorithm designed to control malware propagation and extract network topology information in unknown computer networks. Simulations on a virtual copy of a real intelligent building network in Salamanca demonstrate the algorithm's ability to discover all network nodes and adapt to network characteristics.