Spatial computing applications increasingly require trajectory-level input data to capture how objects move through space. While origin–destination (OD) flows are widely used and interpretable, OD-only data are insufficient for applications that rely on detailed path information. At the same time, raw trajectory data are highly sensitive, as they can reveal individuals’ mobility patterns, which makes data sharing and public release challenging. This paper proposes a method for releasing synthetic mobility trajectories under differential privacy by treating the OD flow as the target workload. Instead of releasing original trajectories, we first compute OD flows from private data and apply a differential privacy mechanism to obtain a privacy-preserving OD target. A synthetic trajectory dataset is then generated and released to satisfy this target. By confining differential privacy to the OD workload, the proposed approach enables the release of trajectory-level microdata while providing formal user-level privacy guarantees. Experimental results show that the proposed method reliably follows the privatized OD tar- gets across different privacy budgets and achieves lower OD error compared to baseline methods, demonstrating its effectiveness for privacy-preserving trajectory data release in spatial computing applications.
Jong-Wook Kim (Sat,) studied this question.