Enabling Federated Process Mining within a Multi-Tier Supply Chain : By integrating differential privacy

Global supply chains moved over $24 trillion worth of goods in 2022, facilitating the delivery of essential products and services worldwide. These supply chains span multiple countries and involve many organisations, making them increasingly complex and interdependent. Due to this complexity, it has become more important than ever for companies to work together and leverage data sharing to gain full visibility across tiers. Operational success can no longer be achieved alone. Federated Process Mining (FPM) was developed to support this collaboration across organisations, without the need to share sensitive data. While this approach presents immense opportunities for supply chain management, its practical adoption is still limited due to persistent privacy concerns. This study addresses this gap by exploring how Differential Privacy (DP), one of the most powerful definitions of privacy, can be integrated into FPM without compromising analytical accuracy. Specifically, the Laplace mechanism is applied to lead time data in a simulated multi-tier supply chain, based on both real and synthetic datasets. The analysis applied a hyperparameter grid search to examine how different privacy levels (𝜀), sample sizes (number of observations), and data variance levels affected the inherent privacy-accuracy trade-off within DP in the context of lead time correlations. The accuracy was measured with the Mean Absolute Error (MAE). The results reveal that stronger privacy (ε< 1.0) inevitably leads to higher accuracy loss, but moderate privacy levels (ε> 1.0) and larger datasets can mitigate these losses. This highlights the need for careful parameter tuning to maintain accuracy under privacy constraints. This study advances the practical implementation of privacy-preserving FPM by addressing the privacy issues and quantifying the privacy-accuracy trade-off under different parameter settings. The results provide both theoretical and practical insights, paving the way for secure, collaborative supply chain analytics across organisational boundaries.