Privacy-Preserving Data Aggregation in Vehicular Ad Hoc Networks Using Deep Learning Techniques
T.S. Balaji, Raju Hariharan, Glaret Subin P, S. Balaji, S. Prasanna Bharathi · 2024
Vehicular Ad Hoc Networks (VANETs) are a significant breakthrough in intelligent transportation systems which allow vehicles to exchange information and with other vehicles and the surrounding infrastructure in realtime. But the utilization of vehicular data collection and aggregation poses problem areas in the issues of privacy, security and performance. This research presents a new conceptual model to employ deep learning algorithms for privacy preservation of data aggregated in V ANETs. We adopt differential privacy for providing privacy-preserving capabilities, federated learning and the usage of the blockchain make it more scalable and secure and edge computing to retain efficiency for solving these challenges. It's about adding controlled noise in order to prevent data points from being reverse engineered, differential privacy is used. Federated learning is used to decentralize the data processing, where participation of the central location is opted out and raw data transmission is kept to the barest minimum as a means of ensuring data privacy. Blockchain technology adds another layer of more secure and reliable aggregation of data improving the overall security of transactions and the prevention of alteration or sabotage. Edge computing is incorporated to ensure the real-time response of the system hence minimizing the latency of the system besides enhancing the performance of the system. In this regard, the study demonstrates that the proposed system ideally implements all the three objectives concerning privacy, data usefulness and real-time capability. Although using differential privacy slightly affects the accuracy of a model it greatly enhances the privacy of data. Together with edge computing, federated learning avoids performance degradation in privacy-preserving computations by decentralizing computations and minimizing transmission costs. The combination of block-chains with the existing networks helps to eliminate cases of data manipulation while the minor adverse impacts towards the systems” network latency are packet.