To improve the security of the home Internet of Things (IoT) systems, and solve the problems of lacking unified authentication and authorization mechanism and poor security in the existing research , this paper puts forward an intrusion detection mechanism based on software defined network (SDN). Firstly, in the authentication and authorization part, an authentication scheme based on elliptic curve cryptography (ECC) algorithm and a service authorization scheme based on OAuth protocol are presented to reduce the computing pressure of IoT devices while ensuring the versatility of the security framework. Then, in the intrusion detection part, a deep learning-based intrusion detection scheme is proposed. Specifically, an intrusion detection module is designed and implemented by combining the flow table statistics function of the SDN controller with the deep learning model. Finally, a Floodlight controller-based intrusion detection and defense system platform is designed, and the home Internet of Things environment is simulated by the Mininet simulation tool. The correctness and effectiveness of the proposed intrusion detection and defense mechanism are evaluated and verified through simulation analysis.