Multi‐objective prairie dog optimization algorithm for IoT‐based intrusion detection

Detecting unauthorized access, unusual activities, and data is significant for the security of IoT networks as it helps identify malfunctioning, faults, and intrusions. Intrusion detection methods analyze network information to identify potential misuse or intrusion attacks. This research proposes a multi‐objective prairie dog optimization algorithm (MPDA) to improve its ability to deal with feature selection problems. The proposed algorithm is modified by incorporating the concepts of an archive, grid, and non‐dominance. An archive and a grid are used to save intermediate best results and improve the diversity, respectively. The non‐dominance concept is employed to deal with multiple objectives. On the NSL‐KDD, CIC‐IDS2017, and IoTID20 datasets, MPDA achieves fewer features, higher accuracy, and lower false alarm rates. MPDA outperforms simple classifiers and state‐of‐art multiobjective optimization algorithms in intrusion detection.