Intrusion Detection in Industrial Internet of Things Based on Granular-Ball Intuitionistic Fuzzy Sets

With the deep integration of the Industrial Internet of Things (IIoT) into critical domains such as intelligent manufacturing and energy management, its cybersecurity risks are increasing. Traditional intrusion detection algorithms struggle to effectively handle the heterogeneity, sparsity, and complexity of intrusion patterns in IIoT systems, leading to performance limitations and the need for improved detection capabilities through new algorithms. Therefore, this paper proposes an intrusion detection framework based on Granular-Ball Intuitionistic Fuzzy Sets (GBIFS). The proposed framework introduces a novel class-wise granular-ball generation method integrated into intuitionistic fuzzy sets, improving intrusion pattern analysis by combining the adaptive multi-granularity representation of granular-ball theory with the capability of intuitionistic fuzzy sets to handle uncertainty. In this framework, the proposed generation method is used to construct Granular-Ball Intuitionistic Fuzzy Patterns (GBIFP) that conform to the feature distribution of IIoT data, and then an improved intuitionistic fuzzy distance metric is introduced to achieve precise classification between normal traffic and attack behavior. Extensive experiments on IIoT intrusion detection datasets (e.g., X-IIoTID, TON-IOT, WUSTL-IIOT) and classical network intrusion detection datasets (e.g., KDDCUP99, NSL-KDD, UNSW-NB15) demonstrate the superior performance of the proposed framework under heterogeneous and sparse data conditions. The GBIFS framework proposed in this paper significantly enhances the accuracy and efficiency of intrusion detection, providing a scalable and robust solution for IIoT cybersecurity. Code is available athttps://github.com/QzEylsia7/Intrusion-Detection-using-Granular-Ball-Intuitionistic-Fuzzy-Sets