The existing security methods cannot be applied to the Internet of Things due to the defects of communication nodes. To solve this problem, an intrusion detection model based on repeated game theory is presented. A repeated game model algorithm for detecting malicious nodes is built, and the algorithm of Quantal Response Equilibrium (QRE) is used for optimizing the model and making results reach the Nash equilibrium. Moreover, a common punishment strategy is introduced to improve the success of transfer data in this model. The results of the simulation represent that this model can restrain malicious nodes attacking effectively and improve the efficiency of network.