Solving Urban Network Security Games: Learning Platform, Benchmark, and Challenge for AI Research

TL;DR

This paper introduces GraphChase, an open-source platform for solving Urban Network Security Games (UNSGs), which model interactions among multiple pursuers and an evader on urban road graphs under varying information and communication constraints. GraphChase provides a unified, modular environment (Game, Agent, Solver) built on Gymnasium to enable rapid development, training, and benchmarking of learning-based strategies, including PSRO-based methods and established baselines like CFR-MIX, NSG-NFSP, and NSGZero. The authors demonstrate that GraphChase achieves comparable or faster wall-clock convergence than original implementations and scales to large, city-scale graphs, while exposing scalability limitations of existing methods through extensive experiments on grids and real-world maps (Singapore and Manhattan). The platform serves as a standardized testbed for evaluating equilibrium concepts (NE, TMECom, TMECor) and coordination strategies in multiplayer security settings, with implications for advancing scalable algorithms in general multiplayer games and urban security applications.

Abstract

After the great achievement of solving two-player zero-sum games, more and more AI researchers focus on solving multiplayer games. To facilitate the development of designing efficient learning algorithms for solving multiplayer games, we propose a multiplayer game platform for solving Urban Network Security Games (\textbf{UNSG}) that model real-world scenarios. That is, preventing criminal activity is a highly significant responsibility assigned to police officers in cities, and police officers have to allocate their limited security resources to interdict the escaping criminal when a crime takes place in a city. This interaction between multiple police officers and the escaping criminal can be modeled as a UNSG. The variants of UNSGs can model different real-world settings, e.g., whether real-time information is available or not, and whether police officers can communicate or not. The main challenges of solving this game include the large size of the game and the co-existence of cooperation and competition. While previous efforts have been made to tackle UNSGs, they have been hampered by performance and scalability issues. Therefore, we propose an open-source UNSG platform (\textbf{GraphChase}) for designing efficient learning algorithms for solving UNSGs. Specifically, GraphChase offers a unified and flexible game environment for modeling various variants of UNSGs, supporting the development, testing, and benchmarking of algorithms. We believe that GraphChase not only facilitates the development of efficient algorithms for solving real-world problems but also paves the way for significant advancements in algorithmic development for solving general multiplayer games.

Figures (8)

• Figure 1: The blueprint of our GraphChase platform.
• Figure 2: The core structure and workflow of GraphChase.
• Figure 3: The training procedure on the easy game with a caught probability of 1.
• Figure 4: The training procedure on the hard game with a caught probability of 0.5.
• Figure 5: UNSGs of $7 \times 7$ with the caught probability of 0.5 (left) or 1 (right).