Secure Access Control for Multi-Tenant Cloud Systems Using a Behavioral Zero-Trust Framework with Machine Learning and Blockchain-Based Auditing

Abstract

The expansion and emergence of multi-tenant cloud platforms have posed significant challenges for securing and maintaining reliable access control, especially when considering how users behave at different times and how cyber threats are constantly evolving. Role-based access control and attribute-based access (ABAC) are examples of traditional access control models that fail to be flexible enough to identify and block unauthorized activities and cross-tenant attacks in current cloud settings. To address these limitations, this study suggests a Secure Behavioral Zero-Trust Access Control Framework (SB-ZTAC), which establishes the integration of 3 modules in one unified framework: Machine learning based behavioral analytics, Zero-trust policy enforcement, and Blockchain based auditing. While the framework uses a behavioural- anomaly detection to produce risk-based access decisions, a zero-trust mechanism dynamically applies access policies according to pre-defined thresholds. Furthermore, a blockchain-based audit layer provides secure and tamper-proof access capability events. Multiple real-world cybersecurity datasets, such as the LANL authentication dataset, UNSW-NB15 and CICIDS2017, are utilized to assess the effectiveness of the proposed framework. Experimental results clearly show that the proposed XGBoost can achieve good detection performance with F1-scores near to 98.0% on LANL and 95.5% on UNSW-NB15 and up to 99.9% on CICIDS2017 data set, while its inference latency is low, typically from 0.002–0.005 ms per sample. In addition, there is minimal computation overhead of the audit logging, making it a real-time solution. The proposed approach provides a scalable and practical solution to improve tenant isolation, improve access control and develop trust in cloud environments by providing multi-tenancy security.