Indoor Positioning Privacy Protection Method Based on Federated Learning in MEC Environment

Abstract

Under the current background, it is very important to study the key technologies of new power system edge-to-side security protection for massive heterogeneous power IoT terminals and edge IoT agents, including defense technologies at the levels of device ontology security, communication interaction security, and secure access. Meaning. The new power system edge-to-side security protection technology has a summary impact on the privacy protection of indoor positioning. This paper proposes an indoor positioning privacy protection method based on federated learning in Mobile Edge. Computing (MEC) environment. Firstly, we analyze the learning mechanisms of horizontal, vertical, and transfer-federated learning, respectively, and mathematically describe it based on the applicability of horizontal and vertical-federated learning under different sample data characteristics. Then, the risk of data leakage when data are used for research or analysis is greatly reduced by introducing differential privacy. In addition, considering the positioning performance, privacy protection, and resource overhead, we further propose an indoor positioning privacy protection model based on federated learning and corresponding algorithms in MEC environment. Finally, through simulation experiments, the proposed algorithm and other three algorithms are, respectively, compared and analyzed in the case of two identical datasets. The experimental results show that the convergence speed, localization time consumption, and localization accuracy of the proposed algorithm are all optimal. Moreover, its final positioning accuracy is about 94%, the average positioning time is 250 ms, and the performance is better than the other three comparison algorithms.