Detection of Data imbalance in MANET network based on ADSY-AEAMBi-LSTM with DBO Feature selection

Abstract

<p>A Mobile Ad Hoc Network (MANET) is a temporary wireless network formed by mobile nodes. These nodes cooperate to relay information in a multi-hop fashion, but some malicious nodes can disrupt the network by providing false routing information. Traditional firewalls and encryption methods can’t keep up with the increasing diversity of network threats. To address these issues, Intrusion Detection Systems (IDS) have been developed. In this paper, a new intrusion detection framework named ADSY-AEAMBi-LSTM is introduced. This acronym stands for a bidirectional Long Short-Term Memory (LSTM) model and an adaptive synthetic auto-encoder attention mechanism. The Dung Beetle Optimizer is used to identify optimal features after data preprocessing, followed by classification using ADSY-AEAMBi-LSTM. The study evaluates this model using three datasets: CIC-IDS 2017, UNSW-NB15, and WSN-DS.</p>