Privacy-Aware MapReduce Based Multi-Party Secure Skyline Computation

Abstract

Selecting representative objects from a large-scale dataset is an important task for understanding the dataset. Skyline is a popular technique for selecting representative objects from a large dataset. It is obvious that the skyline computation from the collective databases of multiple organizations is more effective than the skyline computed from a database of a single organization. However, due to privacy-awareness, every organization is also concerned about the security and privacy of their data. In this regards, we propose an efficient multi-party secure skyline computation method that computes the skyline on encrypted data and preserves the confidentiality of each party’s database objects. Although several distributed skyline computing methods have been proposed, very few of them consider the data privacy and security issues. However, privacy-preserving multi-party skyline computing techniques are not efficient enough. In our proposed method, we present a secure computation model that is more efficient in comparison with existing privacy-preserving multi-party skyline computation models in terms of computation and communication complexity. In our computation model, we also introduce MapReduce as a distributive, scalable, open-source, cost-effective, and reliable framework to handle multi-party data efficiently.

Keywords

• skyline
• MapReduce
• distributed system
• information security
• order-preserving encryption
• homomorphic encryption
• big data
• data privacy
• semi-honest model
• multi-party computation