Network Intrusion Detection System Using Reptile Search with Whale Optimization Algorithm and Multi Head Attention Long Short Term-Memory in IoT

The Internet of Things (IoT) introduces new technological advancements for the development of diverse significant applications.To address the need for protection against attacks, fraud and network intrusions, the Intrusion Detection System (IDS) has become a crucial component within organizations.However, due to the limited resources of IoT devices, classifying attacks and training the model on large datasets consumes more time.In this research, the proposed approach combines the Reptile Search Algorithm (RSA) and Whale Optimization Algorithm (WOA) with Multi-head attention with Long Short-Term Memory (MHA-LSTM) for effective and accurate IDS classification.Initially, IDS is obtained from CICIDS2017, CICIDS2018 and NSL-KDD.Pre-processing methods like Standard Scaling and Label Encoding are involved in transforming numerical values, segmenting features and adjusting them using mean and standard deviation to reduce sensitivity.The RSA with WOA is involved in enhancing intrusion detection by selecting relevant features and optimizing the detection process efficiently to solve complex optimization problems.The classification combination of MHA-LSTM allows the model to scale effectively to large datasets and maximum complex tasks without compromising the performance and accuracy.The proposed WOA-RSA -MHALSTM technique is evaluated on CICIDS 2017, CICIDS 2018 and NSL-KDD datasets, achieving higher accuracies of 99.997% on CICIDS2017, 99.99% on CICIDS2018 and 99.99% on NSL-KDD datasets, which is more effective than Deep Neural Network (DNN) and LSTM.