Privacy-Enhancing Technologies in Federated Learning: A Systematic Review

Arias R.; Ochoa K. · Lecture Notes in Networks and Systems (2025) · DOI: 10.1007/978-3-031-96380-3_23

Resumen

This research presents a systematic review of the literature on Privacy-Enhancing Technologies (PETs) applied in Federated Learning (FL). From an initial analysis of 154 research papers, a total of 20 studies published between 2019 and 2024 in Q1 journals from the Sco pus and ScienceDirect databases were thoroughly analyzed. The results indicate that the most implemented PETs in FL are Differential Privacy (DP), Homomorphic Encryption (HE), and Secure Multiparty Computation (SMPC). These technologies have proven effective in protecting distributed data in FL systems. However, their implementation presents significant technical challenges, such as computational overhead and reduced model efficiency and accuracy. Moreover, important regulatory barriers were identified, especially regarding compliance with the General Data Protection Regulation (GDPR). Although PETs enhance privacy, the research confirms that they limit scalability and performance in large-scale environments. Future studies should focus on optimizing these technologies to balance privacy and efficiency while ensuring regulatory compliance. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Instituciones: Universidad Autónoma del Perú