一句话:提出黑盒安全测试工具 FISTS,通过探测主机端口状态变化并用异常检测优先排查隐患,检测 SDN 配置更新引入的漏洞。
论文关注软件定义网络(SDN)配置更新可能引入的安全漏洞。现有方法多为基于模型且与特定 SDN 系统耦合,难以适用于工业界常见的专有 SDN 框架。作者提出 FISTS,一种完全黑盒的安全测试工具:在 SDN 重配置前后探测网络中的主机,自动识别对应节点并判定其端口状态变化,再借助异常检测算法对主机数据进行优先级排序,从而在最大化发现漏洞的同时减少需检查的节点数,并避免连续误报带来的无效排查。作者在 300 个基于不同配置场景的新数据集上评估,结果显示 COF 与 HBOS 取得最佳效果(召回率最高达 0.99),KNN 召回率接近且执行时间最短;FISTS 能在检查不到 10% 监控数据的情况下检测出所有存在漏洞的主机。工具已开源并提供在线演示。
We present FISTS, a tool that enables security testing of configuration updates in software-defined networks (SDNs). In contrast to conventional approaches, which are model-based and coupled to a specific SDN system, FISTS is entirely black-box thus enabling testing of different platforms, which is necessary for most industries since they typically use proprietary SDN frameworks. FISTS works by probing the hosts on a network before and after an SDN reconfiguration, automatically identifying corresponding nodes, and determining their port state change; finally, it leverages anomaly detection algorithms to prioritize the inspection of hosts data. FISTS also enable engineers to minimize the number of inspected nodes while maximizing the vulnerabilities found by avoiding the inspection of consecutive false alarms. We have evaluated FISTS on 300 new and unique datasets based on distinct configuration scenarios; our results show that FISTS leads to best results (up to 0.99 recall) with COF and HBOS, but KNN leads to close recall and minimal execution time. Further, it enables detecting all vulnerable hosts by inspecting less than 10% of the monitored data. A demo of FISTS is available online at the following URL: https://youtu.be/UIMwFqAvAXg. The tool is available (open source) at: https://doi.org/10.5281/zenodo.18201382.