ASCON-MNASNET: An Effective Data Privacy and Security Framework in Cloud Environment

Abstract

As cloud computing continues to proliferate, users are becoming more concerned about the security and privacy of their data, particularly in light of the growing incidences and complexity of cyberattacks. Therefore, it has become imperative for both individuals and organizations to implement a privacy-preserving intrusion detection system (IDS) to secure the data and detect intrusions. Previously available methods are often inadequate, as they may not effectively balance the need for robust security with the preservation of user privacy, leading to potential vulnerabilities and a lack of trust among clients. To overcome these obstacles, this article introduces CryptoIDS, a novel privacy-preserving IDS that closely combines deep learning-based attack detection with lightweight cryptography. To protect cloud data privacy, CryptoIDS specifically uses a lightweight encryption technique based on ASCON and a CondenseNet-MNasNet hybrid deep learning model for precise and rapid intrusion detection. The framework was thoroughly tested on three benchmark datasets: Cleveland (for privacy evaluation), BoT-IoT and IoT-23 (for security evaluation). Experimental results show that CryptoIDS obtained high detection accuracies of 99.67% on the BoT-IoT dataset and 99.45% on the IoT-23 dataset and improved encryption performance by over 13.89% when compared to current cryptographic algorithms. These findings establish CryptoIDS as a highly effective solution for enhancing both data security and privacy protection in cloud environments.