A Performance of Computational Intelligence for Security in Wireless Networks

Abstract

Wireless Sensor Networks (WSNs) have been a crucial IoT development and, while strong advantages, security problems remain. New cyberattacks are growing as more computers are linked to the internet, following well-known attacks that represent serious risks to the security, credibility, and efficiency of data in WSNs. For many software and scientific questions, the implementation of intelligent computing works effectively; but the defense systems focused on computational intelligence (CI) are not being adequately examined. In this article, it examined two WSN intrusion detection evolutionary computing strategies. A neural network with backpropagation was connected with a classifier of the support vector machine. The ADFA-LD and ADFA-WD datasets were used to determine the detection rates of cyberattacks attained by the two approaches. According to the study, both strategies provide good intrusion detection solutions with an elevated true positive rate and an extremely small false-positive percent. Additionally, by demonstrating its responsibility to preserve small data sets and, consequently, a reasonable FPR ratio below the appropriate limit, it demonstrates the suitability of neural network classification techniques for identifying abnormalitiese.