边云协同环境下智能家居物联网入侵检测方法

Intrusion Detection Method of Internet of Things for Smart Home in Edge-Cloud Collaborative Environments

针对当前入侵攻击更新迭代速度快导致入侵检测方法面临失效性的问题,提出一种利用在线集成学习的入侵检测模型。该方法不仅能够具备在线增量学习的能力,还具备集成学习的方法,上述两种方法的融合不仅能够满足模型的时效性,还能保证稳定性和可用性,避免系统的预测结果随着训练数据的分布偏差过大而导致模型方差变大的问题。为了满足现在边云协同环境下智能家居物联网入侵检测的要求,利用Hadoop分布式计算系统来完成在线集成模型的部署,保证入侵检测模型的高可用、易扩展,使得在线集成模型可以实时应对物联网环境下大吞吐量数据传输和实时处理的需求。

Aiming at the failure of the current intrusion detection method due to the rapid update iteration of intrusion attacks,an intrusion detection model utilizing online ensemble learning is proposed.The model has the ability of online incremental learning and integrated learning at the same time,which can not only meet the timeliness,but also ensure stability and availability,and avoids increasing the model variance due to the large distribution deviation of the system prediction results and the training data.In order to meet the requirements for intrusion detection of the smart home Internet of Things in the edge-cloud collaborative environment,the online ensemble model is deployed using Hadoop distributed computing system.The high availability and easy scalability of the proposed model are guaranteed,and can handle high-throughput data transmission and processing in the IoT environment in real-time.