This code is the implementation of the paper Engorgio: An Arbitrary-Precision Unbounded-Size Hybrid Encrypted Database via Quantized Fully Homomorphic Encryption.

git 
gcc >= 10
cmake >= 3.16
GMP 6.2.0 

You can build the Engorgio (out-of-source) for your machine by executing the following commands:

mkdir build
cd build
cmake ..
make

Following the instructions above, output binaries will be generated in the build/bin/ directory. You can run these binaries by:

$./bin/comparison_test
$./bin/sort_test
$./bin/sync_test
$./bin/topk_test
$./bin/relational_query_test
$./bin/vectorized_query_test_1
$./bin/vectorized_query_test_2
$./bin/hybrid_query_test_1
$./bin/hybrid_query_test_2

• codes test/comparison_test.cpp
• output binary build/bin/comparison_test
• This demo shows the homomorphic comparison operator ArbQuantComp in Engorgio. Execute the binary file comparison_test, we will get the latencry of Engorgio in fig 5a and fig 5b.

• codes test/sort_test.cpp
• output binary build/bin/sort_test
• This demo shows the homomorphic sorting operator HomSort in Engorgio. Execute the binary file sort_test, we will get the latencry of Engorgio in fig 5c. Although we have reduced the number of experiments, it still takes a long time (20+ hours) to complete the experiment.

• codes test/sync_test.cpp
• output binary build/bin/sync_test
• This demo shows the homomorphic synchronization operator HomSync in Engorgio. Execute the binary file sync_test, we will get the latencry of Engorgio in fig 5d.

• codes test/topk_test.cpp
• output binary build/bin/topk_test
• This demo shows the homomorphic Topk operator HomTopK in Engorgio. Execute the binary file topk_test, we will get the latencry of Engorgio in fig 5e.

• codes test/relational_query_test.cpp
• output binary build/bin/relational_query_test
• This demo evaluates relational queries on an encrypted database with varying numbers of rows. Execute the binary file relational_query_test, we will get the latencry of Engorgio in fig 6a and fig 6b.

• codes test/vectorized_query_test_1.cpp and test/vectorized_query_test_2.cpp
• output binary build/bin/vectorized_query_test_1 and build/bin/vectorized_query_test_2
• This demo evaluates vectorized queries on an encrypted database with varying numbers of rows. Execute the binary files vectorized_query_test_1 and vectorized_query_test_2, we will get the latencry of Engorgio in fig 6c and fig 6d. Although we have reduced the number of experiments, it still takes a long time (20+ hours) to complete the experiment.

• codes test/hybrid_query_test_1.cpp and test/hybrid_query_test_2.cpp
• output binary build/bin/hybrid_query_test_1 and build/bin/hybrid_query_test_2
• This demo evaluates hybrid queries on an encrypted database with varying numbers of rows. Execute the binary files hybrid_query_test_1 and hybrid_query_test_2, we will get the latencry of Engorgio in fig 6e. Although we have reduced the number of experiments, it still takes a long time (20+ hours) to complete the experiment.

@inproceedings{engorgio,
 author = {Song Bian and Haowen Pan and Jiaqi Hu and Zhou Zhang and Yunhao Fu and Jiafeng Hua and Yi Chen and Bo Zhang and Yier Jin and Jin Dong and Zhenyu Guan},
 title = {Engorgio: An Arbitrary-Precision Unbounded-Size Hybrid Encrypted Database via Quantized Fully Homomorphic Encryption},
 booktitle = {{USENIX} Security Symposium ({USENIX} Security)},
 year = {2025}
}