This paper proposes an agent-based simulation to analyze the effective resource allocation strategies for patrolling and inspection in consideration of urban network structures. A model of attackers and defend-ers is first formulated as "security game for urban networks." Then optimal security plans are calculated as Nash equilibria, supposing complete rationality of attackers behaviors. The "rational" security plans should be evaluated under more realistic conditions. So we provide an agent-based attacker model by ad-dressing bounded rationality of human behaviors to support decision making under complexity and uncer-Tainty. In particular, this paper mainly evaluates the effectiveness of security plans from the viewpoint of the structural characteristics of urban network, which can affect the route constrains for attackers. Our simulation shows the success rates of security with two types of urban networks and explains the reason why the results are generated through the investigation of attacker agents' micro behaviors in detail.