infosystemshub

Authors: N. Mathivanan and B. Lanitha

Source: Journal of Cloud Computing, Volume 15, Article 18, 2026

DOI:  https://doi.org/10.1186/s13677-026-00858-w

While cloud migration provides processing flexibility, it simultaneously, leaves enterprises vulnerable to cyberthreats. Addressing this risk, Mathivanan and Lanitha (2026) outline a security architecture pairing algorithmic anomaly tracking with an Extended Zero Trust system. Their core thesis states that historical intrusion models fail inside complex cloud networks since they rely heavily on static signatures and rigid set rules.

This assessment is backed by prior literature highlighted in their manuscript. Kim et al. (2020) demonstrated deep learning’s power elevate network defense, whereas Guezzaz et al. (2021) confirmed that data-quality management allows decision-tree variants to achieve much better reliability. Additionally, Cavusoglu (2019) verified that blending diverse machine-learning models optimizes classification accuracy. Taken together, these foundational texts justify the authors’ choice to use a multi-tiered ensemble configuration instead of risking everything on an isolated algorithm.

Technically, the blueprint coordinates four distinct architectures: Support Vector Machines, Random Forest, Long Short-Term Memory networks, and Autoencoders. Supervised tools act as a directory to categorize documented exploits, while unsupervised components detect unusual behavioural deviations pointing to undocumented zero-day threats. This dual implementation aligns with analytical work by Ali et al. (2022) regarding zero-day identification tools. It also maps directly to Goswami’s (2024) argument that AI-driven anomaly systems can enhance real-time surveillance by catching elusive data variations that legacy firewalls overlook.

A key attribute of this research is that it moves beyond detection to automate active incident response. The Zero Trust engine enforces strict multi-factor verification, micro-segmentation, and role-based permissions. This mirrors Sharma’s (2022) findings that continuous monitoring minimizes an organization’s accessible cloud attack surfaces, alongside Flora’s (2020) strategies for microservice threat containment. By routing real-time threat scores directly into access-control loops, the system instantly revokes user credentials, quarantines compromised nodes, or freezes high-risk accounts.

Empirically, the reported metrics are remarkably high: 98.6% precision, a 1.3% false-alarm rate, and 98.4% threat-mitigation success. The researchers stress-tested their platform using benchmark datasets like UNSW-NB15, TON_IoT, CIC-IDS2018, and CICIoT2023. Such cross-environment validation is highly useful since standalone models frequently fail when introduced to unfamiliar variables. The testing protocol logically extends ensemble classification studies by Khan and Haroon (2023) and Krishnaveni et al. (2021), who proved that smart feature selection effectively reduces false alarms.

Nevertheless, these glowing performance claims lack transparent methodological backing. The manuscript glosses over critical execution details, failing to clearly outline training-testing splits, hyperparameter adjustments, or the precise performance weight of each individual algorithm. This makes independent replication impossible. Furthermore, while the paper claims its framework outperforms tools like Snort and Suricata, it fails to present a detailed side-by-side baseline analysis to justify the claim.

Internal reporting inconsistencies further weaken the study. One section highlights average precision, recall, and F1 scores of 98.3%, 98.9%, and 98.5%, yet Table 4 presents different figures at 95.8%, 93.7%, and 94.7% without any explanation. Crucially, the data-availability statement claims no data was analyzed, flatly contradicting the methodology section detailing the use of four distinct databases.

In closing, the manuscript presents a highly relevant cloud defense architecture. Integrating artificial intelligence with Zero Trust protocols offers a robust conceptual approach backed by foundational literature. Even so, the academic community cannot fully endorse the system’s real-world viability until the authors provide transparent model parameters, reproducible data, clear statistical metrics, and more rigorous comparative testing

References

Ali, S., Rehman, S. U., Imran, A., Adeem, G., Iqbal, Z., & Kim, K. I. (2022). Comparative evaluation of AI-based techniques for zero-day attacks detection. Electronics, 11(23), 3934. https://doi.org/10.3390/electronics11233934

Cavusoglu, U. (2019). A new hybrid approach for intrusion detection using machine learning methods. Applied Intelligence, 49, 2735–2761. https://doi.org/10.1007/s10489-019-01405-0

Flora, J. (2020). Improving the security of microservice systems by detecting and tolerating intrusions. In Proceedings of the IEEE International Symposium on Software Reliability Engineering Workshops (pp. 131–134). https://doi.org/10.1109/ISSREW51248.2020.00062

Goswami, M. J. (2024). AI-based anomaly detection for real-time cybersecurity. International Journal of Research and Review Techniques, 3(1).

Guezzaz, A., Benkirane, S., Azrour, M., & Khurram, S. (2021). A reliable network intrusion detection approach using decision tree with enhanced data quality. Security and Communication Networks. https://doi.org/10.1155/2021/709579910.1155/2021/7095799

Khan, M., & Haroon, M. (2023). Detecting network intrusion in cloud environment through ensemble learning and feature selection approach. SN Computer Science.

Kim, H., Kim, H., Kim, Y., & Kim, H. (2020). A deep learning-based intrusion detection method for security enhancement in cloud computing. IEEE Access, 8, 123066–123078.

Krishnaveni, S., Sivamohan, S., Sridhar, S. S., & Prabakaran, S. (2021). Efficient feature selection and classification through ensemble method for network intrusion detection on cloud computing. Cluster Computing, 24(3), 1761–1779. https://doi.org/10.1007/s10586-020-03253-0

Mathivanan, N., & Lanitha, B. (2026). An integrated methodology for intrusion detection and mitigation to optimize cloud security. Journal of Cloud Computing, 15, Article 18.

Sharma, H. (2022). Zero trust in the cloud: Implementing zero trust architecture for enhanced cloud security. ESP Journal of Engineering and Technology Advances, 2(2), 78–91.