buptsseGJ/IoTSeeker

IoTSeeker

I. Introduction

It's a function semantic learning based cross-platform binary vulnerability search approach for IoT devices.

II. File functionality

• 1_gen_idb_file.py: it generates the idb files of binary based on IDA Pro.
• 2_gen_features.py: it extracts original instruction features of basic blocks from idb files.
• 3_gen_train_dataset.py: it generates the training dataset.
• 4_gen_tfrecord_iotseeker.py: it transforms both the instruction features, graph representations of samples into a input file in the tfrecord format, which is used to train the model.
• 5_train_by_list_iotseeker.py: it trains the model based on the input file in the tfrecord format.
• 6_gen_search_tfrecord_iotseeker.py: it transforms both the instruction features, graph representations of samples containing the vulnerability into a input file in the tfrecord format.
• 7_search_by_list_iotseeker.py/7_search_model_iotseeker.py: it conducts the known vulnerability search for binaries across compilation scenarios.

III. Usage

1. We need modify the config.py file. All the dependency directories can be modified here. Simple modification is listed as following, but it need to follow the directory structure we defined:

IDA32_DIR = "installation directory of 32-bit IDA Pro program"
IDA64_DIR = "installation directory of 64-bit IDA Pro program"

2. We put the programs to be searched in the 0_Libs/search_program directory.
3. We run the command.py file to generate the labeled semantic flow graphs and extract initial numerical vectors for basic blocks. The result files should be placed in the 1_Features/search_program directory.
4. We execute the search_by_list_iotseeker.py file to obtain embedding vectors of the functions and get the function list in descending order of similarity scores.