Enhancing cloud data security with ECC-SVM: a machine learning-driven framework for encryption and threat detection
MD Imran Khan1 and Mohan Kumar Patel1
Corresponding Author : MD Imran Khan
Received : 25-November-2025; Revised : 15-December-2025; Accepted : 18-December-2025
Abstract
Cloud computing provides scalable and flexible computing resources but remains vulnerable to unauthorized access, data leakage, and network-based attacks. This study proposes an integrated elliptic curve cryptography-support vector machine (ECC-SVM) framework for strengthening cloud data security through combined cryptographic protection and intelligent threat detection. ECC is employed to encrypt sensitive data before cloud storage, while SVM analyzes incoming network traffic and distinguishes legitimate requests from malicious activities. The framework incorporates user authentication, ECC key generation, data encryption, cloud storage, traffic feature extraction and preprocessing, SVM-based threat classification, secure access control, and ECC decryption. A synthetic cloud security dataset containing 12,000 traffic records was generated, comprising 6,000 legitimate and 6,000 malicious samples representing brute-force login, distributed denial-of-service, unauthorized access, SQL injection, and port-scanning attacks. Twenty traffic and behavioral features were considered, and the dataset was divided into 80% training and 20% testing subsets. Experimental results demonstrate that the proposed ECC–SVM framework achieves 98.75% accuracy, 98.61% precision, 98.83% recall, and a 98.72% F1-score. The receiver operating characteristic analysis produced an area under the curve of 0.988, demonstrating strong discrimination between legitimate and malicious traffic. Batch-wise evaluation further demonstrated stable classification performance, with accuracy ranging from 98.54% to 99.00%. Cryptographic evaluation showed that ECC encryption and decryption times increased gradually with data size, reaching 0.286 s and 0.221 s, respectively, for 20 MB data. Overall, the results demonstrate the potential of the proposed framework to provide integrated data confidentiality and intelligent threat detection for secure cloud computing environments.
Keywords
Cloud computing, Elliptic curve cryptography, Support vector machine, Cloud security, Intrusion detection, Data encryption.
Cite this article
Khan MI, Patel MK. Enhancing cloud data security with ECC-SVM: a machine learning-driven framework for encryption and threat detection. ACCENTS Transactions on Information Security. 2025;10(37):1-11. DOI : 10.19101/TIS.2024.935004