Singamaneni Krishnapriya
Biography
Singamaneni Krishnapriya is an Assistant Professor in the Department of Computer Science and Engineering (CSE ? Cyber Security, Data Science & AI) at VNR Vignana Jyothi Institute of Engineering and Technology (VNR VJIET), Hyderabad. With over 13 years of teaching experience, she is an active researcher in the fields of artificial intelligence, cybersecurity, and Internet of Things (IoT). Her work primarily focuses on deep learning for intrusion detection, federated learning for privacy preservation, blockchain-based attack mitigation, and Quality of Experience (QoE) optimization in smart environments. She has published in reputed journals and international conferences and is passionate about guiding student innovation in cutting-edge domains like edge AI and intelligent network security. Known for her adaptability and commitment to continuous learning, Krishnapriya actively contributes to institutional research initiatives, hackathons, and government-funded proposal development in emerging technologies.
Research Interest
Cybersecurity, Internet of Things (IoT), Deep Learning, Federated Learning, Blockchain, Network Traffic Analysis.
Abstract
Federated Learning for Privacy-Preserving Cybersecurity in Distributed IoT Environments
As cyber threats continue to evolve in sophistication, traditional centralized cybersecurity solutions struggle to cope with the scale, diversity, and privacy demands of modern distributed networks. This work explores the application of Federated Learning (FL) as a privacy-preserving paradigm for cybersecurity in heterogeneous IoT and edge environments. FL enables decentralized entities to collaboratively train intrusion detection and threat classification models without sharing raw data, thereby preserving user privacy and ensuring regulatory compliance (e.g., GDPR, HIPAA).
We propose a federated intrusion detection framework that leverages deep neural networks with differential privacy and secure aggregation to mitigate adversarial threats while maintaining data confidentiality. Extensive experiments conducted on benchmark IoT datasets (e.g., TON_IoT, CICIDS2018) demonstrate that the federated models achieve near-centralized accuracy (94?96%) while significantly reducing privacy leakage risk. Furthermore, we evaluate the system?s robustness against poisoning attacks and communication overhead in realistic edge scenarios.
This study highlights the viability of federated learning as a resilient, scalable, and privacy-preserving cybersecurity approach, opening future directions in federated threat intelligence sharing, secure model aggregation, and adaptive defense mechanisms for 6G and smart city infrastructures.
