---

Securing Data Analytics on SGX with Randomization

Details can be found in here.

Project:

1. App/App.cpp contains the untrusted code.
2. Enclave/Analytics/ contain the data analytics used in various experiments in the paper.

Settings (in App/App.cpp):

1. basedir: Directory of input data files.
2. Experiments settings on chunk/minibatch size, proportion of dummy data instances, number of clusters in K-Means clustering.
3. Class label prediction analytics include Decision Tree, Naive Bayes and K-means. Each classifier can be run by choosing appropriate code, as given in App.cpp under ocall_manager function.

---

How to Build/Execute the Code

1. Install Intel(R) SGX SDK for Linux* OS
2. Specify data directory in basedir of App/App.cpp.
3. Also specify appropriate settings in App/App.cpp file.
4. Build the project with the prepared Makefile: a. Hardware Mode, Debug build: $ make SGX_MODE=HW SGX_DEBUG=1 b. Hardware Mode, Pre-release build: $ make SGX_MODE=HW SGX_PRERELEASE=1 c. Hardware Mode, Release build: $ make SGX_MODE=HW d. Simulation Mode, Debug build: $ make SGX_DEBUG=1 e. Simulation Mode, Pre-release build: $ make SGX_PRERELEASE=1 f. Simulation Mode, Release build: $ make
5. Execute the binary directly: $ ./app