ebook img

Coding Theory lecture: MDS codes PDF

0.06 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Coding Theory lecture: MDS codes

Department of Mathemati s, Mahidol University Kit Tyabandha, PhD MDS ode th 20 January 2006 Theorem 1. Given a redundan y r and a minimum distan e d. An [n;n(cid:0)r;d℄- ode satis(cid:12)es d(cid:20)r+1. x De(cid:12)nition 1. A linear [n;k;d℄ ode over F with d = n(cid:0)k +1 is alled a maximum distan e separable (MDS) ode. In other words, an MDS is a [n;n(cid:0)r;r+1℄- ode. x Theorem 2. Suppose 2(cid:20)r (cid:20)q. Let a1;:::;aq(cid:0)1 be the non-zero elements of GF(q). Then the matrix 2 3 1 1 (cid:1)(cid:1)(cid:1) 1 1 0 (cid:1)(cid:1)(cid:1) 0 66 a1 a2 (cid:1)(cid:1)(cid:1) aq(cid:0)1 0 1 (cid:1)(cid:1)(cid:1) 077 H =66 a21 a22 (cid:1)(cid:1)(cid:1) a2q(cid:0)1 0 0 ::: 077 64 ... ... ... ... ... ... ...75 r(cid:0)1 r(cid:0)1 r(cid:0)1 a1 a2 (cid:1)(cid:1)(cid:1) aq(cid:0)1 0 (cid:1)(cid:1)(cid:1) 1 is the parity he k matrix of an MDS q+1;q+1(cid:0)r;r+1 ode. Equivalently, the olumns of H form a (q+1)-ar in PG(r(cid:0)1;q). x Theorem3. LetC bealinear[n;k;d℄ odeovera(cid:12)eldF ofq elements,whereq isaprimepower with a parity he k matrix H. Then C has a ode word of weight w (cid:20)l if and only if l olumns of H are linearly dependent. x Theorem 4. Let C be a linear [n;k;d℄ ode overF with a parity he k matrix H. Then C is an MDS ode if and only if every n(cid:0)k olumns of H are linearly independent. x ? Theorem 5. If a linear [n;k;d℄ ode C is MDS, then so is its dual C . x Corollary 5[1℄. LetC bean[n;k;d℄linear odeoverF =GF(q). Thenthefollowingstatements are equivalent. a. C is MDS b. Every k olumns of a generator matrix G of C are linearly independent . Every n(cid:0)k olumns of a parity he k matrix H of C are linearly independent x Problem 1. Showthatlinear[n;1;n℄,[n;n(cid:0)1;2℄and [n;n;1℄ odesexist overany(cid:12)nite (cid:12)eldF. x De(cid:12)nition 2. We all trivial MDS odes the [n;1;n℄, [n;n(cid:0)1;2℄ and [n;n;1℄ odes. x Theorem 6. The only binary MDS odes are the trivial ones. x De(cid:12)nition 3. A squarematrixissaid to be non-singularif its olumnsarelinearlyindependent. Given any matrix A, a s(cid:2)s square submatrix of A is a s(cid:2)s matrix onsisting of the entries from some s rows and s olumn of A. x Theorem 7. Let C be an [n;k;(cid:0)(cid:0)℄ ode with parity he k matrix H =(A In(cid:0)k). Then C is an MDS ode if and only if every square submatrix of A is non-singular. x Theorem 8. Let C be an [n;k;(cid:0)(cid:0)℄ ode with generator matrix G = (Ik A). Then C is an MDS ode if and only if every square submatrix of A is non-singular. th Coding theory, MDS ode {1{ From 8 November 2005, as of 20 January, 2006 Department of Mathemati s, Mahidol University Kit Tyabandha, PhD x Theorem 9. Let C be an [n;k;d℄ MDS ode. Then any k symbols of the ode words may be taken as message symbols. x Theorem 10. Let C be an [n;k;d℄ ode over GF(q). Then C is an MDS ode if and only if C has a minimum distan e ode word with non-zero entries in any d oordinates. x Corollary 10[1℄. The number of ode words of weight n(cid:0)k+1 in an [n;k;d℄ MDS ode over GF(q) is (cid:18) (cid:19) n (q(cid:0)1) n(cid:0)k+1 x Problem 2. Given k and q, (cid:12)nd the largest value, m(k;q), of n su h that [n;k;n(cid:0)k+1℄ MDS ode exists overGF(q). x Be ause of Theorem 5, Problem 2 is equivalent to Problem 3. Problem3. Givenk andq, (cid:12)ndthelargestnforwhi hthereisak(cid:2)nmatrixoverGF(q), every k olumns of whi h are linearly independent. x Problem 4. Given a k-dimensional ve tor spa e V over GF(q), what is the order of a largest subset of V every k ve tors of whi h form a basis of the same?. x Theorem 11. For any prime power q, we have m(2;q)=q+1. x Theorem 12. m(k;q)=k+1 for q (cid:20)k. x Bibliography Raymond Hill. A (cid:12)rst ourse in oding theory. Clarendon, 1986 L R Vermani. Elements of algebrai odng theory. Chapman & Hall, 1996 th Coding theory, MDS ode {2{ From 8 November 2005, as of 20 January, 2006

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.