LSPP: a leakage-resilient security approach for a cloud-assisted big data

Data collection has exponentially increased due to the need for new insights in data-centric perspective. This leads to an increased need for scalable and efficient storage and processing technologies. Cloud-based Big Data processing can help businesses and organizations achieve efficient operational performance. However, this massive amounts of data contain sensitive and identifiable personal information, raising concerns regarding their security and privacy in cloud environment. Public cloud-based services continue to face a range of security and privacy challenges in the context of data processing and analysis. The vast amount of data containing sensitive personal information raises security and privacy concerns in public cloud-based services, especially in data processing and analysis. These challenges encompass data breaches and the potential for information leakage because of unauthorized access. Unauthorized entities may gain access to private information by exploiting vulnerabilities in the storage or processing of data or by observing the computation outcome. Therefore, it is crucial to develop an approach that enhances data security and ensures the preservation of privacy during both services’ storage and processing phases. In addition, these services’ utility, effectiveness, and efficiency need to be maintained. Thus, the security approach needs to support the main characteristics of Big Data. To solve these challenges, this paper presents a new security approach that enhances the security of stored and processed cloud-assisted Big Data. By leveraging functional encryption and local differential privacy, a leakage-resilient hybrid cryptographic (LSPP) scheme with valuable privacy and utility is formulated. The LSPP allows performing computation on encrypted samples of the stored data into independent running cloud nodes, which can help ensure high performance. In addition, a calibrated noise is embedded in the computation result to allow a trade-off between privacy and utility. Furthermore, we design a trust quantification for users based on a weighted moving average-OWA operators (WMA-OWA) combination function, which helps prevent inference attacks and improve the system’s efficiency by reducing communication overhead. Detailed security analysis and experimental evaluation are conducted to demonstrate that the LSPP possesses desirable security, utility, and efficiency features. The LSPP does not only balance the utility in noise-sensitive cases, and more significantly, it balances all cloud-based Big Data processing requirements.