On Sunday, June 29, 2025, Professor Dr. Diaa Jassim Kazem, Assistant Dean for Scientific and Graduate Affairs at the College of Engineering, University of Baghdad, chaired the defense committee for the master’s thesis of student Hiba Dhirar Mahmoud from the Department of Information and Communication Engineering at Al-Khwarizmi College of Engineering. The thesis is titled “Federated Deep Learning for Intrusion Detection Systems in Software-Defined IoT Networks.”
The study aims to develop an intelligent and flexible system for detecting intrusions in Internet of Things (IoT) networks, in light of the growing cyber threats and increasing reliance on such networks across various critical sectors.
The researcher built the system using a multi-controller architecture within a Software-Defined Networking (SDN) environment, integrating Federated Learning techniques with deep learning models such as DNN, RNN, LSTM, and CNN. This integration seeks to enhance security and improve performance efficiency.
The system also incorporates the IPFS protocol for secure model storage, along with the AES-256 encryption algorithm to secure communications and ensure data privacy. The system was evaluated using advanced datasets and achieved an accuracy of over 98% with a response time of no more than 20 milliseconds, demonstrating high efficiency in detecting cyberattacks within IoT environments.
