A novel detection and classification framework called the Scaled Hippo Rhinopithecus Swarm Optimized K-three layer Nearest Neighbor model has attained a 98.8 percent classification accuracy and an average security rate of 98.4 percent across four benchmark intrusion datasets. According to the experimental evaluations, the newly developed architecture outperformed existing security models by selectively identifying malicious network events and safeguarding legitimate user records.
Cloud service users and providers regularly encounter operational hurdles related to security threats, data loss, service availability, unauthorized access, and data confidentiality. The study notes that existing security techniques and their associated solutions remain resource-intensive or prove inefficient in delivering reliable cybersecurity protection across cloud environments. The proposed framework was designed to mitigate these computational and protective shortcomings through selective processing.
To assess performance across diverse network scenarios, the evaluation utilized four standard benchmark collections: the UNSW-NB15, CIC-IDS-2017, CSE-CIC-IDS2018, and DARPA datasets. The researchers preprocessed these records through a structured sequence that included data cleaning, handling missing values, data normalization, and dimensionality reduction executed through Scaled Principal Component Analysis.
For optimal feature selection, the architecture incorporates a Hippo Rhinopithecus Swarm Optimization algorithm. This optimization strategy combines the global exploration capability of Hippopotamus Optimization with the local exploitation capability of Rhinopithecus Swarm Optimization. The combined approach isolates critical feature subsets from the preprocessed network records prior to final event classification.
The traffic is subsequently classified using a K-three-layer Nearest Neighbor classifier that implements a majority voting rule. Network traffic is partitioned into sensitive data, defined as legitimate user records requiring confidentiality, and non-sensitive data, representing malicious or attack traffic. To minimize encryption, storage, and computational overhead, the non-sensitive attack traffic is discarded immediately after detection, leaving only sensitive legitimate user data to be encrypted through an Efficient Time-oriented Data Encryption scheme and stored in the cloud.
By selectively encrypting confidential records and discarding attack traffic, the framework optimizes resource utilization while resolving key protection issues faced by cloud users and providers. While the authors indicate that the concept is applicable for real-time intrusion detection, they emphasize that the current experimental evaluation remains limited to offline standard benchmark datasets.
Subject of Research: Cloud intrusion detection and sensitive data encryption using the SHRSO-K3NN framework
Article Title: Efficient detection and encryption of sensitive data in cloud systems using SHRSO-K3NN framework
Article References: Ponnuviji, N. P., Indra Priyadharshini, S., Kandan, M., & Vedaraj, M. (2026). Efficient detection and encryption of sensitive data in cloud systems using SHRSO-K3NN framework. Knowledge and Information Systems, 68(1), Article 313. https://doi.org/10.1007/s10115-026-02924-w
Image Credits: AI Generated
DOI: 10.1007/s10115-026-02924-w
Keywords: SHRSO-K3NN framework, cloud security, intrusion detection, sensitive data encryption, UNSW-NB15, CIC-IDS-2017, CSE-CIC-IDS2018, DARPA dataset, Hippo Rhinopithecus Swarm Optimization, Hippopotamus Optimization, Rhinopithecus Swarm Optimization, K-three-layer Nearest Neighbor, Scaled Principal Component Analysis, Efficient Time-oriented Data Encryption, network traffic classification, feature selection, majority voting rule, offline benchmark evaluation
News Source: Denise Maddox. (November 14, 2026). SHRSO-K3NN Framework Detects Cloud Intrusions With 98.8 Percent Accuracy in Offline Tests. Scienmag.



