A Zero-Trust Software-Defined Networking Architecture for Secure and Trustworthy AIoT Environments

Abstract

This paper describes a Zero-Trust Software-Defined Networking (ZT-SDN) architecture, which has been developed to create secure and trustworthy AIoT environments and provide enhanced network performance overall. The innovative architecture proposed in this paper uses the combination of deep learning-based anomaly detection, trust-aware adaptive routing and continuously-verifying mechanisms to manage network traffic securely and in real time. This is achieved through the continuous monitoring of network activity, real-time detection of malicious activity, and the dynamic updating of SDN flow rules based upon trust evaluation. Experimental results prove the effectiveness of the developed model as compared to traditional methods of machine-learning based approaches in that it achieved an accuracy rate in the area of attack detection greater than 12%, for total of 96.8%. Additionally, using this method reduced packet loss by 30%, lowered the amount of delay (network latency) under heavy (high volume) traffic conditions by 21%, and increased throughput by 18%. The combination of Zero Trust enforcement and Trust Aware Routing provides a means of ensuring the security of transmitted data, as well as mitigating adverse effects from compromised nodes. Thus, the developed Zero Trust SDN Framework provides the user with greater levels of security; scalability; and efficiency, and as a result, is very suitable for implementation in smart cities and AIoT applications for industries.