Abstract: The relentless growth of the internet, coupled with the unprecedented surge in e-commerce activities due to factors such as the global COVID-19 pandemic, has created an expansive digital landscape. However, this flourishing environment has attracted a commensurate increase in cyber threats, particularly concerning the theft of sensitive user information, such as credit card data from e-commerce platforms. This paper introduces an innovative approach by developing a sophisticated Deep Belief Neural Network (DBNN) for intrusion detection which was implemented using Python. This DBNN is seamlessly integrated with Snort, a renowned intrusion detection system, and fortified by the inclusion of a web application firewall. Snort boasts of a robust signature database which aided the identification and elimination of intrusions. A web application firewall is included to foil intrusions at the application layer using rules targeting SQL injection and DoS attacks. By so doing, sensitive customer information such as credit card information which has been a shortcoming with previous systems can be protected. A correlation coefficient of 0.78 between the latency and response time for the baseline and attacked states of the server shows the web application firewall’s ability to maintain the smooth running of the server during intrusion attempts through DoS attacks. In rigorous testing, the DBNN demonstrates a commendable 91.2% accuracy, affirming its efficacy in identifying and thwarting intrusion attempts. The study contributes significantly to knowledge by showcasing that this integrated defense strategy substantially enhances the security posture of e-commerce platforms. A significantly low false positive rate of 8.14% buttresses the effectiveness of the hybrid system in the face of evolving cyber threats in the contemporary digital landscape.
Keywords: Intrusion Prevention System, Deep Belief Neural Network, Denial of Service, E-commerce
|
DOI:
10.17148/IJARCCE.2024.13709
