To effectively provide an early alarm of dangers for attack events, security organizations have successfully employed Intrusion Detection System (IDS) to transfer the suspicious connections to honeypot which can capture and analyze the hacker's behavior and virus signature for years. Using the minefield strategy to deploy honeypot systems, managers place decoy systems and spread them among network nodes to trap hackers. There exists the problem that honeypot constantly cannot appeal the attention of hacker's attack if honeypot is deployed within the inappropriate zone or node. It is a crucial issue that how to effectively deploy it for accumulating large numbers of information as well as decrease the anti-detect possibilities by hackers. Hence, we develop a network-based analysis model for dynamic honeypot deployment through the use of probability theorem and traffic analysis technique to improve the limitations of way of static strategy, promote the decision quality of honeypot deployment. It discovers the best route with the minimum cost, and decides the optimal deployment node to increase the trap possibility within distinct QoS constraints. Using NS2, this model is validated by four network deployment strategies, that is, minimum-cost deployment, random deployment, Bayes-based deployment and dynamic deployment, to test its efficiency. The experimental results show that the proposed approach can effectively locate the recommended nodes and the optimal node of honeypot deployment in a communication network.