This preview shows page 1. Sign up to view the full content.
Unformatted text preview: the code! 33 Rennes presentation Example: DES DPA Signal
Round 1 Round 2 Round 3 DPA Signal KcW
N = 300
100 MS/s
PCW average 34 Rennes presentation Same DES DPA Signal  Zoom
DPA Signal KcW
PCW average Target Bit = 17 (S5) SBOX
DPA Signal 20 17 18 19 PPerm
DPA Signal N = 300
100 MS/s S8 S7 S6 S5 S4 S3 S2 S1
35 Rennes presentation Reverse engineering the code
Round 1 Round 2 XOR Exch RL EPerm SBOX Round 3 XOR Exch RL DPA Signal KcW
N = 300
100 MS/s
PCW average 36 Rennes presentation Brief Platform Description
Oscilloscope
file transfer
Arm scope
retrieve file Scope trigger
on IO Current waveform
acquisition
Server
stores files
and runs Treatment
software Main PC
runs Acquisition
software R
GCR Card extention command emission
Card
reader Protection box 37 Rennes presentation A popular protection against DPA
s Curve desynchronization smears DPA peaks
s Huge number of raw curves required (50000) to do a DPA 38 Rennes presentation The Problem (as seen by the bad guy)
Desynchronization sources can be discrete or continuous :
• quantum binomial random events (dummy cycles = DS)
• voluntary unstable continuous clock drifts
Can we get rid of their impact ?
• Restore the true DPA peaks in a rigorous mathematical way
• Reduce the number of acquisitions
Mathematically :
• Centrallimit theorem è DS addup to a normal law
• The desynchronization’s effect is a convolution product :
measuredDPApeaks = trueDPApeaks«G(T,σ) ò X(t)=a Xtrue (tτ)× Exp[(τT)2/ (2σ 2)] dτ
39 Rennes presentation A look at the scope (unstable clock drift)
Desynchronized Synchronized 1 GS/s
40 Rennes presentation Related tools and problems in physics
s How to repair a blurred image? x As frustrating as it seems, the useful data necessary
to repair a blurred image is already in the image.
x Photons were simply “mixedup” and sent to the wrong
places. s Question: If the blurring function K is known (say the
exact non voluntary movement of the photographer)
can we still restore the image? 41 Rennes presentation A few (spectacular) examples
function K = blurred image = = deconvolved images
(RichardsonLu...
View
Full
Document
This note was uploaded on 10/18/2010 for the course MATH CS 301 taught by Professor Aliulger during the Fall '10 term at Koç University.
 Fall '10
 ALIULGER
 Cryptography

Click to edit the document details