基于多层神经网络的智能家居入侵检测方法

依托物联网技术的智能家居面临多重信息安全风险，现有智能家居入侵检测方案存在难以处理大量高维度数据、检测率低、误检率高、依赖经验确定网络层数等问题。提出一种融合深度学习与模糊神经网络的多层神经网络入侵检测方法；基于深度学习完成数据特征的学习，将高维数据映射为低维数据；基于网络重构误差训练并优化确定网络深度。仿真测试结果表明，该方案可有效提高对攻击行为的检测准确率和检测效率；针对远程非法访问的检测率可达到94%，对拒绝服务攻击的检测准确率可达96%，对网络中新型攻击的检测率超过60%。

Intrusion detection method for smart home based on multilayer neural network

Smart home, relying on Internet of things technologies, is facing multiple risks in information security. The existing solutions have difficulties in handling a large number of high-dimensional data and have problems such as low detection rates, and high rates of false positives, or the number of network layers determined by experience. An intrusion detection method based on multilayer neural network fused deep learning and fuzzy neural network is proposed. The learning of data features is completed by means of deep learning, the high-dimensional data is mapped into low-dimensional one. The network depth is optimally determined by training based on the network reconfiguration error. The simulation test results show that the proposed scheme can effectively improve the detection accuracy and detection efficiency of the attack behavior. The detection rate of remote illegal access can reach 94%, the detection accuracy rate for denial of service (DoS) attacks can reach 96%, and the detection rate of emerging attacks from the network exceeds 60%.