Cyber Threat Prediction Model for Transferring Data in Wireless Edge Computing Platforms Using Deep Learning

Abstract

Wireless edge computing or WEC which is becoming more popular to transmit computation-intensive projects while improving end users' quality of service. Edge devices may access cloud services and functionality via WEC. However, owing to the current spike in attack activity, the expansion of the Internet of Things (IoT) aims poses serious cyber security issues. Hardware and deep learning solutions are being developed to detect cyber-attacks, data dumping and traffic situations in edge networks. Deep models can outperform their shallow counterparts in terms of learning due to the massive amounts of data generated by IoT devices the application of DL models finds assistance in several fields, with the WEC serving as the approach's clear benefactor for traffic forecasting and attack detection. To address these issues, this study offers a unique DL-based traffic forecast approach that combines a cyber-attack prediction strategy with an information transmission mechanism. The suggested approach consists of three basic procedures, intrusion detection, data unloading and traffic forecasting.  In this research, we concentrate on intrusion detection. Initially, we preprocessed the data using the Standard Scalar technique. Principal Component Analysis (PCA) is utilized in feature extraction to minimize dimensionality while preserving important information. Less relevant characteristics are progressively removed for feature selection using recursive feature elimination (RFE). Finally, we proposed a novel Rectified Hyperbolic Long Short-Term Layered ConvoNeuronet (RH-LSTM-CNN) approach for intrusion detection. The suggested strategy outperforms the current approaches with accuracy (98.33%), recall (98%), precision (98%), and f1 score (98%).