Ddos Detection In Software Defined Network Using Federated Learning

Abstract

Distributed Denial Of Service Attacks Have Become A Great Concern For Security In Software Defined Networking(SDN), As They Mostly Overload Centralized Security Mechanisms. In This Work, A Federated Learning-Based Intrusion Detection System(FL-IDS) Using Convolutional Neural Networks(CNN) And Long Short TermMemory(LSTM) Networks Is Proposed. Clients Train CNN-LSTM Models Locally On Network Traffic, Preserving Data Privacy. The Federated Server Aggregates These Models Securely, Using Differential Privacy Techniques. The Trained Global Model Is Then Deployed In SDN Switches To Analyze Real-time Traffic, With Packets Classified According To Specific Attributes: Size, Protocol Type, And Time Intervals. Once An Attack Is Detected, The System Policy On The SDN Switch Is Updated So That Threats Will Be Mitigated Dynamically. By Decentralizing Intrusion Detection, This Approach Increases Accuracy While Protecting Sensitive Data.

Keywords