Multi-user Searchable Attribute Based Encryption for Outsourced Big Data

Petabytes of data are generated every minute, and this creates a corresponding demand for servers which ingest, process, store, and maintain this data. The scale of data has become sufficiently large that normal file access and processing present different challenges than in prior decades. In the Web 2.0 era, data owners commonly stored and shared data using on-premise servers. In the past decade, cloud computing has quickly become the dominant hosting paradigm, as it offers data owners the benefits of cost-savings, flexible scaling, and ease of administration. However, cloud computing presents a myriad of security challenges. Attribute based encryption has been key to tackling these challenges. We present a method to allow users to maintain their own secure data on a shared system, while eliminating the time penalty and single point-of-failure associated with traditional authorization techniques. This is implemented as a Representational State Transfer API, removing the burden of managing auxiliary files or encryption libraries when uploading documents. This system also supports data encrypted at rest, removing concerns of administrator access to unencrypted contents in storage. We leverage key policy attribute-based encryption (KP-ABE) to store this data, search its contents, and receive/decrypt this data. We further demonstrate that KP-ABE can be implemented as part of a distributed or centralized authorization method. We compare KP-ABE against AES cryptography, and present system metrics from its execution on representative data.