This paper introduces an efficient and scalable cloud-based privacy preserving model using a new optimal cryptography scheme for anomaly detection in large-scale sensor data. Our proposed privacy preserving model has maintained a better tradeoff between reliability and scalability of the cloud computing resources by means of detecting anomalies from the encrypted data. Conventional data analysis methods have used complex and large numerical computations for the anomaly detection. Also, a single key used by the symmetric key cryptographic scheme to encrypt and decrypt the data has faced large computational complexity because the multiple users can access the original data simultaneously using this single shared secret key. Hence, a classical public key encryption technique called RSA is adopted to perform encryption and decryption of secure data using different key pairs. Furthermore, the random generation of public keys in RSA is controlled in the proposed cloud-based privacy preserving model through optimizing a public key using a new hybrid local pollination-based grey wolf optimizer (LPGWO) algorithm. For ease of convenience, a single private server handling the organization data within a collaborative public cloud data center when requiring the decryption of secure sensor data are allowed to decrypt the optimal secure data using LPGWO-based RSA optimal cryptographic scheme. The data encrypted using an optimal cryptographic scheme are then encouraged to perform data clustering computations in collaborative public servers of cloud platform using Neutrosophic c-Means Clustering (NCM) algorithm. Hence, this NCM algorithm mainly focuses for data partitioning and classification of anomalies. Experimental validation was conducted using four datasets obtained from Intel laboratory having publicly available sensor data. The experimental outcomes have proved the efficiency of the proposed framework in providing data privacy with high anomaly detection accuracy.

中文翻译：

---

云环境中异常检测的最优密码方案和高效中智C-Means聚类

本文介绍了一种高效且可扩展的基于云的隐私保护模型，该模型使用一种新的最优密码方案，用于大规模传感器数据中的异常检测。我们提出的隐私保护模型通过检测加密数据中的异常情况，在云计算资源的可靠性和可扩展性之间保持了更好的平衡。传统的数据分析方法使用复杂且大量的数值计算来进行异常检测。此外，对称密钥加密方案用于加密和解密数据的单个密钥面临着巨大的计算复杂性，因为多个用户可以使用这个单个共享密钥同时访问原始数据。因此，采用称为 RSA 的经典公钥加密技术，使用不同的密钥对对安全数据进行加密和解密。此外，RSA 中公钥的随机生成在所提出的基于云的隐私保护模型中通过使用新的混合本地授粉灰狼优化器 (LPGWO) 算法优化公钥来控制。为方便起见，当需要解密安全​​传感器数据时，允许在协作公共云数据中心内处理组织数据的单个私有服务器使用基于 LPGWO 的 RSA 最佳加密方案解密最佳安全数据。然后鼓励使用最佳密码方案加密的数据在云平台的协作公共服务器中使用中智c-均值聚类（NCM）算法进行数据聚类计算。因此，该 NCM 算法主要侧重于异常的数据划分和分类。使用从英特尔实验室获得的具有公开可用传感器数据的四个数据集进行实验验证。实验结果证明了所提出的框架在提供具有高异常检测精度的数据隐私方面的效率。