Optimizing General-purpose Compiler for Private Distributed Computation

Description:

This technology details a means of achieving secure computing of sensitive information in a multi-party, shared resource environment. The computing of sensitive information can now be accomplished in a cloud computing framework without risking data security. Applications exist in the health, financial, and other information-sensitive markets or when sensitive data is shared between separate organizations.

The private distributed computation compiler (PICCO) is a source-to-source complier that works on any platform and translates a program written in an extension of the C programming language to its secure distributed implementation in C. Before translation, data may be annotated as private, and the resulting distribution can be subsequently compiled and securely outsourced. PICCO is flexible, giving each programmer the option to direct whether operations are executed in parallel or sequentially, in order to achieve maximum performance depending on the task. Additional performance enhancement is gained by allowing array operations including private data.

The data flow remains secure through several measures. Restrictions are put in place when private and public data are both being used; most notably, preventing the assignment of private values to public variables. Similarly, conditional statements with private conditions are not allowed to contain observable public actions. Memory access at private locations is also masked to protect the location information.

Patent Information:
For Information, Contact:
Tim Joyce PH
Licensing Associate
University of Notre Dame
 
Inventors:
Marina Blanton
Yihua Zhang
Keywords:
© 2017. All Rights Reserved. Powered by Inteum