0% found this document useful (0 votes)
28 views

Stream Ciphers: PRG Security Defs

This document summarizes key concepts from an online cryptography course taught by Dan Boneh. It introduces pseudorandom generators (PRGs) and defines what it means for a PRG to be secure and indistinguishable from random. A secure PRG should have negligible advantage over any efficient statistical test. It is shown that a PRG is secure if it is unpredictable, meaning no efficient algorithm can predict its next outputs with non-negligible probability. More generally, two distributions are computationally indistinguishable if no efficient algorithm can tell them apart.

Uploaded by

Boo
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
28 views

Stream Ciphers: PRG Security Defs

This document summarizes key concepts from an online cryptography course taught by Dan Boneh. It introduces pseudorandom generators (PRGs) and defines what it means for a PRG to be secure and indistinguishable from random. A secure PRG should have negligible advantage over any efficient statistical test. It is shown that a PRG is secure if it is unpredictable, meaning no efficient algorithm can predict its next outputs with non-negligible probability. More generally, two distributions are computationally indistinguishable if no efficient algorithm can tell them apart.

Uploaded by

Boo
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

Online

  Cryptography   Course                                                                             Dan   Boneh  

Stream  ciphers  

PRG  Security  Defs  

Dan  Boneh  
n
Let      G:K  ⟶  {0,1}      be  a  PRG    
 
Goal:        define  what  it  means  that  
 
   

 is  “indisHnguishable”  from  

Dan  Boneh  
StaHsHcal  Tests  
Sta$s$cal  test  on  {0,1}n:        
   an  alg.    A    s.t.      A(x)    outputs    “0”  or  “1”  
 

Examples:  
 

Dan  Boneh  
StaHsHcal  Tests  
More  examples:  

Dan  Boneh  
Advantage  
Let      G:K  ⟶{0,1}n      be  a  PRG        and        A    a  stat.  test  on    {0,1}n  
 
Define:      
 
 
 
 
 
A  silly  example:        A(x)  =  0      ⇒        AdvPRG  [A,G]  =        0   Dan  Boneh  
Suppose    G:K  ⟶{0,1}n    saHsfies      msb(G(k))  =  1        for  2/3  of  keys  in  K  

Define  stat.  test    A(x)    as:  


 if    [    msb(x)=1    ]    output  “1”  else  output  “0”  

Then  

 AdvPRG  [A,G]    =    |  Pr[  A(G(k))=1]    -­‐    Pr[  A(r)=1  ]  |    =    

         |  2/3  –  1/2  |  =      1/6  

Dan  Boneh  
Secure  PRGs:        crypto  definiHon  
Def:      We  say  that      G:K  ⟶{0,1}n      is  a  secure  PRG  if      

Are  there  provably  secure  PRGs?  


 but  we  have  heurisHc  candidates.    
Dan  Boneh  
Easy  fact:          a  secure  PRG  is  unpredictable  
We  show:          PRG  predictable      ⇒      PRG  is  insecure  
 
 
Suppose    A    is  an  efficient  algorithm  s.t.  
 
 
   
   for  non-­‐negligible    ε        (e.g.      ε  =  1/1000)  

Dan  Boneh  
Easy  fact:          a  secure  PRG  is  unpredictable  
Define  staHsHcal  test    B    as:  

Dan  Boneh  
Thm  (Yao’82):          an  unpredictable  PRG  is  secure  
Let    G:K  ⟶{0,1}n    be    PRG  
 
“Thm”:          if      ∀  i  ∈  {0,  …  ,  n-­‐1}    PRG    G    is  unpredictable  at  pos.    i  
           then        G    is  a  secure  PRG.  
 
 
If    next-­‐bit  predictors  cannot  disHnguish  G  from  random  
 then  no  staHsHcal  test  can  !!  
 
Dan  Boneh  
Let    G:K  ⟶{0,1}n      be  a  PRG  such  that    
 from  the  last  n/2  bits  of  G(k)    
 it  is  easy  to  compute  the  first  n/2  bits.  
 
Is    G    predictable  for  some  i  ∈  {0,  …  ,  n-­‐1}    ?  

Yes  
No  
More  Generally  
Let      P1      and      P2      be  two  distribuHons  over    {0,1}n  
 

Def:        We  say  that  P1  and  P2  are    


     computa$onally  indis$nguishable    (denoted                                      )  
 
 
 

R  
Example:      a  PRG  is  secure  if      {  k  ⟵K  :    G(k)  }    ≈p    uniform({0,1}n)  
Dan  Boneh  
End  of  Segment  

Dan  Boneh  

You might also like