13.07.2015 Views

Applications of finite geometry in coding theory and cryptography

Applications of finite geometry in coding theory and cryptography

Applications of finite geometry in coding theory and cryptography

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Construction 1Let PG(V ) be an n-dimensional(space with basis e i (0 ≤ i ≤ n).(n+d+1 ) )Let PG(W) be <strong>and</strong>+1 − 1 -dimensional space with basis e i0,...,i d(0 ≤ i 0 ≤i 1 ≤ · · · ≤ i d ≤ n).To simplify notations, we will write e i0,...,i dwith 0 ≤ i 0 , . . .,i d ≤ n when we meanthe vector e iσ(0) ,...,i σ(d)where σ is a permutation with 0 ≤ i σ(0) ≤ i σ(1) ≤ · · · ≤ i σ(d) ≤n.Let θ : V d+1 → W be the multil<strong>in</strong>ear mapp<strong>in</strong>g.θ : (n∑i 0=0x (0)i 0e i0 , . . .,n∑i d =0x (d)i de id ) ↦→∑0≤i 0,...,i d ≤nx (0)i 0· . . . · x (d)i de i0,...,i d. (4)For each po<strong>in</strong>t P = [x] <strong>of</strong> PG(V ), we def<strong>in</strong>e a subspace D(P) <strong>of</strong> PG(W) byD(P) = 〈θ(x, v 1 , . . .,v d ) | v 1 , . . . , v d ∈ V 〉 . (5)Theorem 15The set D = {D(P) | P ∈ PG(V )} is a generalised dual arc with dimensions d i =)− 1, i = 0, . . . , d + 1.( n+d+1−id+1−iPro<strong>of</strong>. S<strong>in</strong>ce θ is a multil<strong>in</strong>ear form, we getD(P 0 ) ∩ · · · ∩ D(P k−1 ) = 〈θ(x 0 , . . . , x k−1 , v k , . . .,v d ) | v k , . . . , v d ∈ V 〉<strong>and</strong> hence dim(D(P 0 ) ∩ · · · ∩ D(P k−1 )) = ( )n+d+1−kd+1−k − 1. (The −1 is because theprojective dimension is one less than the vector space dimension).□The l<strong>in</strong>k between dual arcs <strong>and</strong> MACs is:Theorem 16Let π be a hyperplane <strong>of</strong> PG(n + 1, q) <strong>and</strong> let D be a generalised dual arc <strong>of</strong> order l <strong>in</strong>π with parameters (n, d 1 , . . . , d l+1 ).The elements <strong>of</strong> D are the messages <strong>and</strong> the po<strong>in</strong>ts <strong>of</strong> PG(n + 1, q) not <strong>in</strong> π arethe keys. The authentication tag that belongs to a message <strong>and</strong> a key is the generated(d 1 + 1)-dimensional subspace.This def<strong>in</strong>es a perfect MAC <strong>of</strong> order r = l + 1 with attack probabilitiesp i = q di+1−di .Pro<strong>of</strong>. After i message tag pairs (m 1 , t 1 ), . . . , (m i , t i ) are sent, the attacker knows thatthe key must lie <strong>in</strong> the (d i + 1)-dimensional space π = t 1 ∩ · · · ∩ t i . This space conta<strong>in</strong>sq di+1 different keys. A message m i+1 <strong>in</strong>tersects m 1 ∩ · · · ∩ m i <strong>in</strong> a d i+1 -dimensionalspace π ′ . Two keys K <strong>and</strong> ¯K generate the same authentication tag if <strong>and</strong> only if K <strong>and</strong>¯K generate together with π ′ the same (d i+1 +1)-dimensional space. Thus the keys formgroups <strong>of</strong> size q di+1+1 <strong>and</strong> keys from the same group give the same authentication tag.The attacker has to guess a group. The probability to guess the correct group isp i = q di+1+1 /q di+1 .□

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!