MOE KLINNS Lab, Xi'an Jiaotong University, Xi'an 710049, China.
Shenzhen Research School, Xi'an Jiaotong University, Shenzhen 518057, China.
Sensors (Basel). 2019 Feb 24;19(4):958. doi: 10.3390/s19040958.
System security monitoring has become more and more difficult with the ever-growing complexity and dynamicity of the Internet of Things (IoT). In this paper, we develop an Intelligent Maintenance and Lightweight Anomaly Detection System (IMLADS) for efficient security management of the IoT. Firstly, unlike the traditional system use static agents, we employ the mobile agent to perform data collection and analysis, which can automatically transfer to other nodes according to the pre-set monitoring task. The mobility is handled by the mobile agent running platform, which is irrelevant with the node or its operation system. Combined with this technology, we can greatly reduce the number of agents running in the system while increasing the system stability and scalability. Secondly, we design different methods for node level and system level security monitoring. For the node level security monitoring, we develop a lightweight data collection and analysis method which only occupy little local computing resources. For the system level security monitoring, we proposed a parameter calculation method based on sketch, whose computational complexity is constant and irrelevant with the system scale. Finally, we design agents to perform suitable response policies for system maintenance and abnormal behavior control based on the anomaly mining results. The experimental results based on the platform constructed show that the proposed method has lower computational complexity and higher detection accuracy. For the node level monitoring, the time complexity is reduced by 50% with high detection accuracy. For the system level monitoring, the time complexity is about 1 s for parameter calculation in a middle scale IoT network.
随着物联网（IoT）的日益复杂和动态化，系统安全监控变得越来越困难。在本文中，我们开发了一种智能维护和轻量级异常检测系统（IMLADS），用于高效管理物联网的安全。首先，与传统系统使用静态代理不同，我们采用移动代理来执行数据收集和分析，它可以根据预设的监控任务自动转移到其他节点。移动性由移动代理运行平台处理，与节点或其操作系统无关。结合这项技术，我们可以在减少系统中运行代理数量的同时，提高系统的稳定性和可扩展性。其次，我们为节点级和系统级安全监控设计了不同的方法。对于节点级安全监控，我们开发了一种轻量级的数据收集和分析方法，只占用很少的本地计算资源。对于系统级安全监控，我们提出了一种基于草图的参数计算方法，其计算复杂度是常数，与系统规模无关。最后，我们根据异常挖掘结果设计代理，执行适合系统维护和异常行为控制的响应策略。基于构建的平台进行的实验结果表明，所提出的方法具有较低的计算复杂度和较高的检测精度。对于节点级监控，检测精度很高的情况下，时间复杂度降低了 50%。对于系统级监控，在中等规模的物联网网络中，参数计算的时间复杂度约为 1 秒。
