CSR: Small: Privacy and Security for MapReduce Clouds with PASMAC

点击翻译按钮获取中文摘要

This research targets the design and evaluation of protocols forsecure, privacy-preserving data analysis in an untrustedcloud. Therewith, the user can store and query data in the cloud,preserving privacy and integrity of outsourced data and queries. ThePIs specifically address a real-world cloud framework: Google'sprominent MapReduce paradigm.Traditional solutions for single server setups and related work on,e.g., fully homomorphic encryption, are computationally too heavy anduneconomical and offset cloud advantages. The PIs' rationale is todesign new protocols tailored to the specifics of the MapReducecomputing paradigm. The PIs' methodology is twofold. First, the PIsdesign new protocols that allow the cloud user to specify dataanalysis queries for typical operations such as searching, patternmatching or counting. For this, the PIs extend privacy-preservingtechniques, e.g., private information retrieval or order preservingencryption. Second, the PIs design protocols guaranteeing genuinenessof data retrieved from the cloud. Using cryptographic accumulators,users can verify whether data has not been tampered with. Besidesdesign, the PIs also implement a prototype that is usable in arealistic setting with MapReduce.The outcome of this project enables privacy-preserving operations andsecure data storage in a widely-used cloud computing framework, thusremove one major adoption obstacle, and make cloud computing availablefor a larger community.