Optimized neural network model for attack detection in LTE network

Abstract

Machine-type communication (MTC) senses our environment througfh connecting millions of devices to one another, and becomes an enabler for context-aware and ubiquitous computing services. Still, there has been no research managing the network lifetime for these MTC devices due to a lack of authentication requirements and attack detection. Thus, this research work intends to develop an efficient attack detection model for MTC. Since the previous work on MTC based Long Term Evaluation(LTE) model had only authenticated the MTC devices via clustering, security is a critical issue that is focused on the current research work. Here, the attack detection is undergone by introducing an enhanced Neural Network (NN) model, where the authenticated nodes, as well as cluster Head (CH) from elliptic curve cryptography (ECC), are detected for attacks. In case of the presence of attack, the enhanced NN introduces a penalty as one and prohibits those nodes from taking part in MTC communication. As a novelty, here the training of enhanced NN is accomplished via a new optimization algorithm referred to Whale with Tri-level Update (WTU). Moreover, the security requirement is fulfilled by fixing the objectives like confidentiality and repudiation on data transmission. The efficiency of the proposed attack detection model is proved and a comparative evaluation will be accomplished in terms of certain security analysis.