MobiDiff: Semantic-Aware Multi-Channel Discrete Diffusion for Human Mobility Data Generation

2026-07-09

AI Summary

Existing approaches to generating discrete human mobility data that preserve semantic structures—such as regions, activities, and temporal patterns—are often inefficient and rely on complex interpolation or post-processing pipelines. This work proposes MobiDiff, the first end-to-end multi-channel discrete diffusion framework that simultaneously denoises spatial, activity, and temporal semantic channels without requiring latent trajectory construction. Its key innovation lies in a joint masking mechanism operating at the event, population, and channel levels, which effectively captures both global trajectory structure and intra-event dependencies. Experimental results demonstrate that MobiDiff accurately preserves trajectory length and inter-event time interval distributions across the Atlanta, Boston, and Seattle datasets, while achieving an average inference speedup of 5.3× over GeoGen.

Abstract

Human mobility data are essential for transportation optimization, urban planning, and resource allocation, yet real-world mobility data are costly to collect and difficult to share due to privacy concerns. Recent diffusion-based methods have shown promise in synthesizing realistic mobility patterns, but they typically rely on continuous or latent spatio-temporal traces, limiting their ability to natively model discrete semantic events with explicit region, activity, time, and interval structures. To address this issue, we introduce MobiDiff, an end-to-end discrete diffusion framework that efficiently generates mobility data by directly denoising multi-channel semantic skeletons, avoiding the costly interpolation, latent trace construction, and coarse-to-fine realization pipelines widely used in existing diffusion-based methods. Specifically, MobiDiff decomposes each human check-in event into spatial, activity, and temporal channels, and employs structured event-, group-, and channel-level masking to jointly capture trajectory-level mobility patterns and within-event dependencies. We evaluate generation fidelity, privacy-preserving, and efficiency on three large-scale real-world datasets from Atlanta, Boston, and Seattle. Results show that MobiDiff effectively preserves trajectory length and temporal interval distributions while remaining competitive across broader mobility statistics; it is also much faster than state-of-the-art methods, e.g., 5.3$\times$ faster than GeoGen on average during inference. These findings suggest that discrete diffusion offers an interpretable and efficient framework for synthetic mobility data generation.

Problem

• human mobility
• discrete diffusion
• semantic events
• data generation
• privacy

Innovation

• discrete diffusion
• semantic-aware generation
• multi-channel modeling
• human mobility synthesis
• privacy-preserving generation