Hello,
As part of a school project, I am trying to perform a simple power analysis attack on a naïve RSA implementation by exploiting the modular exponentiation algorithm. I run my algorithm on a Joy-IT Arduino UNO R3 DIP (ATmega328P). To measure power consumption, I inserted a 1 Ω shunt resistor in series with the power supply.
Hardware Setup:
External power supply: 5 V voltage generator
Oscilloscope: RIGOL DS1042C (40 MHz, 400 MSa/s)
Measurement method: measuring the voltage across the shunt resistor to deduce current variations
A GPIO (PIN_SYNC) is used as a synchronization signal to trigger the oscilloscope. Basically, the Arduino outputs a continuous 5 V signal while executing the modular exponentiation algorithm. This allows me to know precisely when the Arduino is performing that part (it is clearer when you check out the code).
My objective is to observe temporal variations in power consumption during modular exponentiation, in order to distinguish between square and multiply operations.
Code running on the Arduino:
const int PIN_SYNC = 12;
void setup(){
pinMode(PIN_SYNC, OUTPUT);
digitalWrite(PIN_SYNC, LOW);
}
unsigned long modexp(unsigned long base, unsigned long exposant, unsigned long mod){
unsigned long resultat = 1;
base = base % mod;
while (exposant > 0) {
if (exposant & 1) {
resultat = (resultat * base) % mod;
}
base = (base * base) % mod;
exposant >>= 1;
}
return resultat;
}
void loop(){
digitalWrite(PIN_SYNC, HIGH);
modexp(7, 105, 187); // 105 = 1101001 in binary
digitalWrite(PIN_SYNC, LOW);
delay(1000);
}
Problem: the voltage variation across the shunt resistor is very small, close to the noise, and I struggle to clearly distinguish the operations in the power trace.
I am therefore wondering:
Is 1 Ω too small for this type of attack on an Arduino? Maybe I am having a shunt issue. I tried to take a higher shunt but it wasn’t effective.
Is it realistic to expect exploitable observations without signal amplification or current sensor?
Is a 40 MHz oscilloscope sufficient for this type of measurement?
Do you have any practical advice (shunt placement, AC/DC coupling, filtering, clock frequency, etc.) to improve my setup?
I would greatly appreciate any feedback or suggestions. I have added in attachment a picture of my circuit (in french sorry).
Thank you in advance.