Development Of Machine Learning Based Security Algorithm For 4g Network Against Wormhole.

International Journal of Research and Innovation in Applied Science (IJRIAS) | Volume VII, Issue II, February 2022 | ISSN 2454–6194

Development Of Machine Learning Based Security Algorithm For 4g Network Against Wormhole.

¹Eze E.M., ²Ituma C., ³Asogwa T.C., ⁴Ebere U.C.
123Enugu State University of Science and Technology
4Destinet Smart Technologies Ltd.

Abstract
This work development of machine learning based security algorithm for 4G network against wormhole. This was achieved using methods such as data collection, data extraction, training and classification process. The system design employed mathematical and structural method to develop the models of the wormhole and also the new security algorithm using artificial neural network. This was implemented with Simulink and neural network toolbox, before testing. The result showed that the algorithm was able to detect wormhole at a regression of 0.9978 and Mean square error of 2.05×10^-5. The security algorithm deployed on a 4G network and tested; the result showed that throughput percentage of 89.16% and latency of 76.325ms which according to International Telecommunication Union (ITU-U) and Nigerian Communication Commission standard (NCC) are good.

Keywords: Wormhole, 4G Networks, Security, Machine Learning, Neural Network

I. INTRODUCITON

Over the years, the rapid growth in Internet of Things (IoTs) connectivity has equally presented the need for information technology (IT) companies to upgrade their capacity and network size so as to meet up with the demand for data and management of user equipments. To achieve this aim, the various IT companies have constantly upgrade their service equipments with the transformation of the broadband ecosystem from the original state of 3G to the presents day state of the art Long Term Evolution (LTE) network. The LTE is today know as 4G network with proposed solution to wireless broadband and heterogeneous network connectivity; providing increase bit rate of over 20Mbps and improved service quality (Payaswini et al., 2013). This advancement in the IT sector has lead to the rapid adoption of the technology; as today internet has become part of man with limitless applications. According to Contel (2020), security is ranked as both the primary benefit and biggest challenge for IT professionals. This is to say that despite the huge benefit the internet service offers, addressing the security concerns is the only way to take its full advantage. It is therefore imperative to carefully analyze these security issues and the conventional security approaches proposed and implemented in the past with a view to improve the performance using a better solution.
Over the past decades, these issues of information security in wireless network have been one of the most active research area, with (Kolias et al., 2016; Khosroshahy et al., 2013) identifying some of the attack trends as man in the middle, wormhole, botnet, denial of service, black hole, spoofing among other attack forms to mention a few. In these attack types mentioned, all take different forms and requires different approaches for mitigation, however the wormhole type is very dynamics and till date still remain the main attack tool for hackers to bring down network nodes. According to Nicklas (2017) wormhole attack is one of the serious security issues currently faced by wireless networks. The wormhole attack (WHA) creates an illusion of two nodes which can attract large amount of network traffic and as a result manipulates the network to launch series of attacks. To solve this problem various techniques have been proposed such as neighbor discovery algorithm, encryption algorithm and machine learning, but all have their limitations; however the use of machine learning provided the most reliable security solution for wormhole. Machine learning (ML) is intelligent system which has the ability learn from pattern recognition problems and make accurate decisions (Mehta, 2019). This involves many algorithms such as support vector machine, K-nearest neighbor, artificial neural network (ANN) among others, however the effectiveness of ANN in other fields when compared to other ML counterparts have made it special to solve this problem and will be used to train clusters of wormhole vector and then deploy on the 4G network for security.

Read More