Abstract:Industrial Cyber-Physical Systems (CPSs) require flexible and tolerant communication networks to overcome commonly occurring security problems and denial-of-service such as links failure and networks congestion that might be due to direct or indirect network attacks. In this work, we take advantage of Software-defined networking (SDN) as an important networking paradigm that provide real-time fault resilience since it is capable of global network visibility and programmability. We consider OpenFlow as an SDN protocol that enables interaction between the SDN controller and forwarding plane of network devices. We employ multiple machine learning algorithms to enhance the decision making in the SDN controller. Integrating machine learning with network resilience solutions can effectively address the challenge of predicting and classifying network traffic and thus, providing real-time network resilience and higher security level. The aim is to address network resilience by proposing an intelligent recommender system that recommends paths in real-time based on predicting link failures and network congestions. We use statistical data of the network such as link propagation delay, the number of packets/bytes received and transmitted by each OpenFlow switch on a specific port. Different state-of-art machine learning models has been implemented such as logistic regression, K-nearest neighbors, support vector machine, and decision tree to train these models in normal state, links failure and congestion conditions. The models are evaluated on the Mininet emulation testbed and provide accuracies ranging from around 91-99% on the test data. The machine learning model with the highest accuracy is utilized in the intelligent recommender system of the SDN controller which helps in selecting resilient paths to achieve a better security and quality-of-service in the network. This real-time recommender system helps the controller to take reactive measures to improve network resilience and security by avoiding faulty paths during path discovery and establishment.

FASHION: Functional and Attack graph Secured HybrId Optimization of virtualized Networks

Vulnerabilities and Attacks of Inter-device Coordination in Programmable Networks.

FASHION: Fault-Aware Self-Healing Intelligent On-chip Network

Enhancing Network Resilience through Machine Learning-powered Graph Combinatorial Optimization: Applications in Cyber Defense and Information Diffusion

Effective DDoS Mitigation Via ML-Driven In-network Traffic Shaping

Dynamic Network Security Function Enforcement via Joint Flow and Function Scheduling

Exploiting the Vulnerability of Flow Table Overflow in Software-Defined Network: Attack Model, Evaluation, and Defense

Securing industrial communication with software-defined networking

Attack Graph Obfuscation

A Firewall Optimization for Threat-Resilient Micro-Segmentation in Power System Networks

A Predictive Framework for Cyber Security Analytics using Attack Graphs

Fashion: A Fast and Accurate Solution to Global Routing Problem

Deployment Optimization of IoT Devices through Attack Graph Analysis

Predictive Cyber-security Analytics Framework: A non-homogenous Markov model for Security Quantification

Extending Attack Graphs to Represent Cyber-Attacks in Communication Protocols and Modern IT Networks

Security Function Virtualization Based Moving Target Defense of SDN-Enabled Smart Grid.

Efficient SFC Protection Method Against Network Attack Risks in Air Traffic Information Networks

Information Flow Coverage Metrics for Hardware Security Verification

The Integration of Protection, Restoration, and Adaptive Flow Redistribution in Building Resilient Networked Critical Infrastructures Against Intentional Attacks

Load-Optimization in Reconfigurable Data-Center Networks: Algorithms and Complexity of Flow Routing

Dependable Virtualized Fabric on Programmable Data Plane