DTDS: Dilithium Dataset for Power Analysis

doi.org/10.57760/sciencedb.j00173.00001DTDS: Supplement to the Dilithium dataset for side channel energy analysis The dataset consists of 60,000 power traces, corresponding metadata, and sensitive intermediate values extracted from the metadata. Among them, 50,000 power traces are generated when random plaintext is encrypted with random private keys (referred to as the modeling dataset), and 10,000 power traces are generated when random plaintext is encrypted with a fixed private key (referred to as the recovery dataset).This dataset contains a total of 36 files with a size of 3 GB, which are stored in 12 subfolders respectively.

The subfolders are named from 000 to 011. Among them, subfolders 000 to 009 are used to store the modeling dataset (i.e., using random private keys), while subfolders 010 to 011 are used to store the recovery dataset (i.e., using a fixed private key). As shown in Table 3, each subfolder contains 3 files, namely the power trace file (named polyz_unpack_traces.npy, with each row representing one power trace, totaling 5,000 traces), the metadata file (named mate.req, totaling 5,000 entries), and the sensitive intermediate value file (polyz_unpack_y.npy, totaling 5,000 entries).The dataset also provides sample code for functions such as dataset reading and visualization, which is stored in the test.py file.