Greedy Quadratic Neural Backed Decision Tree ...

Greedy Quadratic Neural Backed Decision Tree Continuous Authentication for Zero Trust Secure Satellite Communication

Author Name : Varadala Sridhar

ABSTRACT Zero Trust security in satellite communication made an appearance as an intriguing mechanism for addressing pitfalls of conventional network security. Providing information security has become laborious and cumbersome with advancements in information technology and identified practical applications since origin of digital age. Also attack intensity to cyber-security has risen exceedingly over the past few years. Zero Trust security in satellite communication is now in its infancy stage with primary focus on access control, trust evaluation and identity authentication. In this work a novel method called, Incremental Greedy Quadratic Neural Backed Decision Tree Classifier (IGQ-NBDTC) for zero trust secure satellite communication is proposed. First, an improved Multi-facets Zero Trust Architecture (MZTA) that improves conventional zero-trust concept to multi-facet zero-trust by focusing on subject, object, environment and behavior. Then a Six Dimensional Proximity Minority Removal-based Preprocessing to ensure smooth and balanced dataset is designed. Next, an Incremental Greedy Quadratic Programming Feature Selection is designed that usingQuadratic Programming reduces latency and training time in an incremental fashion. In addition the Neural Backed Decision Tree Classifier for Continuous Authentication is also designed to boost the authentication accuracy. Experimental results disclose enhanced authentication accuracy and latency by identification of intruders in a timely mannerusing STIN dataset. To evaluate the IGQ-NBDTC method performance, four different performance metrics, security, latency, authentication accuracy, false positive rate are validated and analyzed. As a result, IGQ-NBDTC method achieved a higher performance compared to other state-of-the-art methods