IoT Intrusion Detection through Traffic Feature Categorization and Preprocessing Optimization

Intrusion Detection System (IDS) has a vital importance in IoT-based security solutions, as IoT networks present distinctive issues due to resource limitations and diverse traffic characteristics. The current paper introduces the design of an IDS model by implementing the feature classification and mathematically consistent pre-processing approaches. With the classification of traffic characteristics into four classes, namely basic connection, content-based, statistical, and traffic volume, together with the relevant preprocessing procedures, the new framework is capable of improving the learning stability and minimizing the learning complexity at the IoT gateway. In addition, a feasible MLPNN-based approach is recommended as the classifier for the proposed solution. While this study is mainly dedicated to the theoretical design and analysis of the IDS framework, implementation and empirical investigation will be considered in further research.