Secure multiparty variance estimation in unbalanced resource environments
Cong Hu, Zhen Yao, Jiali Sun, Cuiling Liu, Cui-Cui Zhang, Ruixuan Lu, YuJia Zhai · 2024
In big data statistical analysis tasks, the diversity of data types and the complexity of computing tasks are both increasing. Distributed computing, which involves processing data across multiple nodes, is widely used but raises security concerns such as data leakage, tampering, and forgery. Hence, it is vital to explore privacy protection in distributed computing environments. This paper investigates secure computation mechanisms suitable for multi-party scenarios characterized by unbalanced computing resources. We integrate local differential privacy with secure multi-party computation technologies to develop a hybrid variance estimation method. The proposed hybrid method weights the estimation results of the local differential privacy algorithm to mitigate its error impact on result accuracy. Furthermore, we design experiments to assess the computing effectiveness of this hybrid method. The results demonstrate that the hybrid variance estimation method achieves a superior balance between accuracy and computational overhead compared to methods relying solely on a single privacy protection technique.