Iobt intrusion detection system using machine learning
Abstract
Internet of Battlefield Things (IoBT) is a granular approach to military operational effectiveness that draws in-spiration from the Internet of Things (IoT) paradigm. Rather than networking home appliances and light fixtures to opti-mize their energy usage, IoBT connects military assets and systems such as combat equipment, personal devices, armored and unmanned vehicles, and sensors. The resulting system is both an information gathering and distribution network that augments battlefield efficiency, autonomy, and real-time decision-making capabilities of the personnel. The modular approach of IoBT is both its biggest advantage and its Achilles' heel. An IoBT can be seamlessly adapted and scaled according to the battlefield needs, but the availability and accuracy of the data shared between nodes is vulnerable to tampering, errors, and hacking. Subsequently, malicious actors can access confidential data, taint it, or prevent parts of IoBT from functioning. To fortify the cybersecurity aspect of IoBT, all involved personnel should maintain the quality of the information, which includes its integrity and confidentiality. To detect intrusion in IoBT, we propose a multi-faceted intrusion detection system that meshes ensemble methods with supervised machine learning to detect and report anomalies. We used CIC-IDS-2017 and CIC-IDS-2018 intrusion datasets for benchmarking classifiers, dividing them into a 70:30 ratio. The performance of the hybrid IDS model is finely tuned to deliver a high detection rate and low false positives rate.