Advanced wireless networking classification using transformer Based inception resnetv2 with non-Linear analysis

Abstract

The fast expansion of wireless networking technologies in recent years has made more sophisticated and accurate classification methods essential to control the increasing complexity and variation of network data. While traditional approaches are successful, their dynamic and non-linear aspect questions many of them. This work addresses these challenges by introducing a new classification system integrating Transformer-based architecture with Inception ResNet V2 enhanced by non-linear analysis methodologies. The need of a model resistant to the inherent non-linearities of wireless network environments and effectively capture both the spatial and temporal dependencies inside the data drives this approach. The proposed method manages the sequential character of wireless data by first strengthened by a Transformer model after Inception ResNet V2 as a feature extractor. By including non-linear analysis, the model can better fit complex patterns that traditional linear models would overlook, therefore enhancing the categorisation process. The classification performance of this hybrid model is evaluated with a big dataset covering several wireless networking environments. Experimental data shows that the proposed Transformer-based Inception ResNet V2 model clearly outperforms traditional machine learning approaches including Decision Trees, KNN, and Neural Networks. The model especially gets an F1 score of 95.3%, a precision of 94.5%, an accuracy of 95.8%, and a recall of 96.2%. Moreover dropped to 3.2% and 2.7% respectively are the false negative and false positive rates (FNR). These results show the effectiveness of combining advanced deep learning architectures with non-linear analysis for wireless networking categorisation, therefore offering a possible way to improve network performance and dependability.