Comprehensive Study on Privacy-Preserving with Artificial Intelligence in Internet of Things

Over the past few years, the Internet of things (IoT) is becoming an essential part of our daily lives by connecting everything from smart home devices to wearable health monitors and industrial sensors to constantly collect and share large amounts of data. This generated massive volumes of real time and heterogeneous data demands intelligent processing and analyzing through Machine Learning (ML) and Deep Learning Techniques (DL) techniques, which enables predictive analysis, automated decision making and contextual awareness. However, several existing ML and DL methods depends on centralized framework, which requires transmission of raw data to the cloud servers, which introduces risks such as unauthorized access, data breach and regulatory non- compliance. This comprehensive study focusses on the urgent need to explore and implement privacy preserving mechanisms, which ensures reliable processing of data within IoT devices. This study examines the solutions such as Federated Learning (FL), Differential Privacy (DP), Secure Multi-party Computation (SMPC), and Homomorphic Encryption (HE), designed to protect the user data by maintaining model accuracy. This study also explores the systems for IoT with string privacy and efficiency, particularly in resource constrained IoT environments. Further, this study discusses the current limitations, reviews the current solutions and highlights the future research and development.