SDN-Based Federated Learning Approach for Satellite-IoT Framework to Enhance Data Security and Privacy in Space Communication

The proliferation of IoT devices and integration of machine learning technologies paved the path towards automation in various sectors such as manufacturing, communication, automobiles, agricultural, health etc. guided by Artificial intelligence (AI). As space exploration is transitioning from a mere idea to a tangible reality, the integration of AI-powered IoT will be an essential aspect of space colonies, where self-governing systems will be the norm. These IoT networks will have a broad range of coverage that will extend to the farthest limits with the aid of low orbit satellite integration. However, the widespread adoption of these IoT technologies is highly contingent on ensuring that data is protected from malevolent intrusions. Therefore, in this paper we have proposed a federated learning based distributed approach in an SDN environment to thwart data breach that can plague satellite-IoT framework with respect to space communication. Additionally, as part of the implementation of the framework, we have devised an SDN backbone equipped with a traffic regulator to prevent malicious traffic flows in the network. Our system correctly classifies malicious traffic, blocks flood sources and ensures safe data transmission between IoT devices. The implemented OpenMined-based federated learning method is promising with an accuracy rate of 79.47% in detecting attacks. Our future work will be focused on improving the accuracy of the federating learning-based approach and in conducting differential privacy-based approaches to demonstrate the privacy related advantages of the proposed framework.