Agent-based modeling (ABM) is a computational simulation method in which individual entities (agents) — such as pedestrians in a crowd — are each governed by local behavioral rules, and macroscopic collective phenomena emerge from the bottom-up interactions among these autonomous agents. In WiFi/CSI-based human sensing and crowd dynamics research, ABM matters because it enables realistic generation of synthetic movement and behavioral data, supports validation of sensing models against ground-truth simulations, and captures fine-grained spatial and temporal heterogeneity that aggregate models obscure. Key variants relevant to the field include cellular automata (CA) models, where agents occupy discrete grid cells with rule-based transitions, hybrid mesoscopic-microscopic frameworks that couple agent-level behavior with population-level kinetic equations, and data-driven ABMs whose transition rules or floor fields are learned directly from empirical trajectory data rather than hand-crafted assumptions.
Source Papers
- A crowd team evacuation model considering spring effect ↗ — A crowd team evacuation model considering spring effect
- A hybrid mesoscopic/agent-based model for crowd dynamics with emotional contagion ↗ — A hybrid mesoscopic/agent-based model for crowd dynamics wit
- Modeling spatial patterns in a moving crowd of people using data-driven approach—A concept of Interplay Floor Field ↗ — Modeling spatial patterns in a moving crowd of people using
- Recent trends in crowd analysis: A review ↗ — Recent trends in crowd analysis: A review
- Social force models for pedestrian traffic – state of the art ↗ — Social force models for pedestrian traffic – state of the ar