Abstract
Numerous factors are causing wireless networks to carry an increasing volume of traffic, which is developing quickly. The 5 G mobile networks target a variety of new used cases that consist of more heterogeneous devices linked with similar frameworks. Moreover, the most frequent and fastest-growing Distributed Denial of Service (DDoS) attack, targets the developing computational network infrastructures worldwide. This makes the development of an effective and early detection for massive, complex DDoS attacks necessary. Therefore, the aim of this study is to present a unique DDoS attack detection model for 5 G networks that consists of two phases: feature extraction and attack detection. Here, Long Short-Term Memory (LSTM) & Recurrent Neural Network (RNN) classifiers are combined to perform the detection. The Opposition Learning-based Seagull Optimization Algorithm (OLSOA) model optimizes the weight of the RNN for better accurate detection. A correctly trained hybrid model produces a detected output that is more accurate. Finally, the outcomes of the adopted strategy are calculated about various metrics using conventional approaches. Particularly, the adopted work's accuracy at node = 10000 outperforms the existing DCNN, RNN, LSTM, Hybrid classifier + WOA, Hybrid classifier + MFO, and Hybrid classifier + SOA methods by 10.8%, 8.33%, 16%, 27.08%, 17.02%, and 5.01%.