Power analysis attack based on the MLP in DPA Contest v4

Power analysis represents extremely effective and successful way of side-channels attacks on so far confidential cryptographic algorithms and cryptographic devices. One of the widespread countermeasures against these power analysis attacks is the masking approach. In 2012, Nassar et al. presented a new lightweight masking countermeasure to protect the AES (Advanced Encryption Standard) implementation. This masking scheme represents the target algorithm of the DPA Contest v4. In this article, we present successful attack based on MLP (Multi-layer Perceptron) aimed on this masking countermeasure. For the first time, the MLP attack is used to reveal secret key from masked implementation of cryptographic algorithm. The article describes in detail every step of the attack implemented including the results achieved. Our attack reveal each byte of the secret key of the masked AES only with 23 power traces.