Low-rate distributed denial of service attacks detection insoftware defined network-enabled internet of things usingmachine learning combined with feature importance

One of the main challenges in developing the internet of things (IoT) is theexistence of availability problems originated from the low-rate distributeddenial of service attacks (LRDDoS). The complexity of IoT makes theLRDDoS hard to detect because the attack flow is performed similarly to theregular traffic. Integration of software defined IoT (SDN-Enabled IoT) isconsidered an alternative solution for overcoming the specified problemthrough a single detection point using machine learning approaches. Thecontroller has a resource limitation for implementing the classificationprocess. Therefore, this paper extends the usage of Feature Importance toreduce the data complexity during the model generation process and choosean appropriate feature for generating an efficient classification model. Theresearch results show that the Gaussian Naïve Bayes (GNB) produced themost effective outcome. GNB performed better than the other algorithmsbecause the feature reduction only selected the independent feature, which hadno relation to the other features.