Feature: Hardware design
Table 1: Hardware fault mitigations
Table 2: Fault injection hardening techniques
If the candidate key is correct, there will
be a clear correlation spike in Δ(t). Tis process is repeated for all possible 256 keys. Te difference values for all the false keys are effectively noise, whilst the value for the true key byte will stand out. Repeat this attack for all 16 key bytes and you have the AES key.
Corelation power analysis A more modern and more powerful form of power analysis attack is called “correlation power analysis (CPA)”. Like DPA, we collect a set of power traces across the same S-BOX operation used for DPA
for a range of known plaintext values. Ten we can use the PC simulation of AES to calculate and store the intermediate values (Vi(k)) for our plaintext data P[i] and an initial candidate key (k):
Intermediate value Vi(k) = S(P[i]⊕k) power trace data Wi[t]
Now, we can use the Hamming weight
of the calculated intermediate values as a leakage power model:
Li(k) = HW(Vi(k))
Equation 1: Pearsons correlation coefficient for two linear values
www.electronicsworld.co.uk February 2026 23
Since the Hamming weight is linear to
the power consumption of the operands, we can use a niſty bit of statistics called Pearsons correlation coefficient, as shown in Equation 1 below. Tis measures the strength and direction (both increasing or decreasing) of a linear relationship between two variables, in this case our leakage model and the real-world recordings.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48
