Applying Feature Engineering to Detect and Classify Attacks in Private Cellular Networks
Open AccessPrivate cellular networks have emerged to address the growing use cases and needs of different industry verticals from manufacturing, healthcare, agriculture, mining, education, campuses and oil fields. Enterprises are being driven by the recent allocation of spectrum by different regulators and benefits such as mobility, control, low latency, reliability, coverage and cost. to build private networks that meet their business, mission-critical use cases and application needs. Deploying these networks in enterprise environments introduces attack vectors and surface in areas such as policy & standards, supply chain and the 5G systems architecture. Enterprises deploying Private Cellular Networks that are not large mobile network operators will likely lack the scale and resources needed to address the multitude of risks associated with the different components of the overall network. They are also faced with high risks of intrusion and threats from the different components that are integrated to build the network. This study examined the use of machine learning feature engineering to detect different attacks accurately and rapidly in the network for mitigation without negatively impacting the performance of the network. To achieve the output, the original data set is from a real 5G network implementation. Various transformations were performed on the dataset to derive new features. Models were then trained and tested with results showing improved performance and predictive accuracy. The research investigated three machine learning models: Generalized Linear Models (GLM), Gradient Boosting Machine (GBM) and Decision Tree. Based on the F1 scores it was determined that LightGBM was good for attack classification and Xgboost was good for attack type classification. These show that derived features can be used to train models for attack detection in Private Cellular Networks deployed in enterprise environments.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.