A significant amount of data has been produced as Internet of Things (IoT) applications are being used in day-by-day activities. IoT applications frequently need for the involvement of several technologies, such as, cloud computing (CC), edge computing (EC), fog computing (FG), green computing (GC), etc. which have created significant security issues. Because of the inefficiency of present security measures, cyberattacks are also increasing as a result of the usage of these technologies. Many artificial intelligence (AI)-based security solutions, including intrusion detection systems, have been created in recent years (IDS). For the creation of smart analytical tools,it is necessary to have preprocessing, feature selection, data augmentation, machine learning (ML) techniques.This study aims to improve intrusion detection accuracy by employing a supervised classification framework and a mode rank-based mayfly optimization algorithm (MRMFOA). It produces the mayfly optimization by redefining the related components to operate on discrete spaces. To be more precise, it introduces a random exploration function that adds additional variety and redefines the concept of distance (between individuals in mayfly optimization). In addition to the random move defined in the MFO algorithm, the latter includes two extra random approaches based on the crossover and mutation operators. On the UNSW-15 datasets, experiments were conducted to assess how well the recommended approach worked. The MRMFOA produced better results by showing 2% of improvement in terms of accuracy.