1 / 35

Parvaresh-Vardy Codes

Explore unique decoding and list decoding algorithms for Reed-Solomon codes, with distinct elements and polynomial encoding. Learn about efficient decoding techniques and interpolation methods. Discover the Parvaresh-Vardy codes and improvements over traditional methods.

tennille
Télécharger la présentation

Parvaresh-Vardy Codes

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Parvaresh-Vardy Codes Derandomization Seminar, 2016 Guy Biron

  2. Reminders

  3. Reed-Solomon Codes • distinct elements of • Message • Polynomial • Encoding of :

  4. Unique Decoding RS – Welch-Berlekamp • Input: pairs with distinct • Find nonzero polynomial and polynomial such that: • Output: or • Corrects up to fraction of errors in polynomial time

  5. List Decoding RS – Definition • Input: received word , error parameter • Output: all polynomials of degree at most such that for at least values of • Should work in polynomial time

  6. List Decoding RS – General Structure • Interpolation step: find non-zero s.t. • Need to ensure number of coefficients is greater than number of constraints • Different implementation in each algorithm, should run in polynomial time • Root finding step: if is a factor of , output (assuming and ) • Need to ensure is a factor of for every required • can be factored in polynomial time (fact)

  7. List Decoding RS – Alg. 1 • Input: , pairs • Find nonzero such that: • For every factor of : • If and : • Output • Corrects up to fraction of errors in polynomial time

  8. List Decoding RS – Alg. 2 (Sudan) • Input: , pairs • Find nonzero such that: • For every factor of : • If and : • Output • Corrects up to fraction of errors in polynomial time

  9. List Decoding RS – Alg. 3 (Guruswami-Sudan) • Input: , pairs • Find nonzero such that: • with multiplicity • For every factor of : • If and : • Output • Corrects up to fraction of errors in polynomial time

  10. Parvaresh-Vardy

  11. Spoiler • Going to have codes and list decoding algorithm that corrects up to fraction of errors in polynomial time ( is constant) • Better than for rates • Key ideas • Instead of bivariate polynomial interpolation, do multivariate interpolation (we will detail the trivariate case ) • Use other (new) codes, not Reed-Solomon

  12. Definitions and Relevant Lemmas

  13. Hasse Derivative • Let be a polynomial over • For , the Hasse derivative of is: • For , the Hasse derivative of at , , is the coefficient of in

  14. Multiplicity • Let and • is said to pass throughwith multiplicity, or to have a zero of multiplicity at this point, if:

  15. Weighted Degree • The weighted degree of is defined by: • The weighted degree of is defined by: • We will define: • The weighted degree of a polynomial is the maximum weighted degree of its monomials

  16. Interpolation • Let be distinct elements of • An interpolation set for is: • The interpolation polynomial with respect to is the least weighted degree nonzero polynomial that has a zero of multiplicity at each of the points of • The interpolation polynomial can be computed in polynomial time • Gaussian elimination, iterative algorithm

  17. Lemma 3 • Let be an interpolation set, be the interpolation polynomial. Then: • Proof: • Number of constraints for passing through points with multiplicity : • Let be the smallest weighted degree such that there surely exists a of weighted degree . Number of unknowns (coefficients): • Demanding , we get the right hand side above

  18. Lemma 4 • Let be an interpolation set, be the interpolation polynomial. Given a pair of polynomials over , define: if and: then

  19. Proof of Lemma 4 • We want to show that • Thus • For each such that , has a zero of multiplicity at . Thus has at least roots. • We have an upper bound for (and thus for ) from lemma 3 • Demanding gives the result

  20. Lemma 5 • satisfies if and only if there exist such that: • Proof: • : obvious • : fix a monomial order s.t. and • can be expressed as: • Polynomial division algorithm guarantees none of the monomials in is divisible by the leading term of or • and , so • , so

  21. Encoding and Decoding Algorithms

  22. Code Parameters • distinct elements of • Multiplicity parameter • Basis for over • Polynomial of degree , irreducible over • Positive integer such that

  23. Encoding Algorithm • Given message , define • Compute over • The codeword corresponding to message is: • Note that the rate of the code is

  24. Encoding Time • and are polynomials of degree at most • Define field . Think of as elements of field • Use and to denote elements of corresponding to and • The computation of becomes in field • Result of Von Zur Gathen and Nöcker shows this is done in polynomial time • Thus encoding the message is done in polynomial time

  25. Decoding Algorithm • Given , decompose each into , where . Set up the interpolation set and compute the interpolation polynomial • Compute • Compute • Find and output roots of

  26. Decoding – Step 1 • . Decompose into , • Set up the interpolation set • Compute the interpolation polynomial

  27. Lemma 6 • Suppose a codeword differs from in at most: positions, and let denote the polynomials that produce . Then • Proof: immediate from lemma 4 by getting a lower bound on • We now know that for “good” messages we get

  28. Decoding – Step 2 • Compute • Write as an element of : • . Let • can be regarded as an element of field • Thus

  29. Lemma 7 • is not the all-zero polynomial. • Proof: • Assume to the contrary that • So for all , meaning is a factor of for all • So is a factor of , and there exists: • is irreducible, so has multiplicity at the points • Furthermore, • Contradiction to being the least weighted degree

  30. Lemma 8 • Suppose , and let denote the elements of corresponding to . Then . • Proof: • By lemma 5, can be written as: • Therefore can be written as: where are the remainders of upon division by • Then • We now know that for “good” messages we get

  31. Decoding – Steps 3,4 • Compute , find and output its roots • Due to the encoding definition, we know that in field • So if , then • We now know that for “good” messages, is a root of • Shoup has shown a polynomial algorithm for finding roots of in polynomial time • So we can find “good” messages in polynomial time! • We just need to show that is not the all-zero polynomial

  32. Lemma 9 • is not the all-zero polynomial • Proof: • Assume to the contrary that • Thus is a factor of • This cannot happen if • Since , we get that • From lemma 3 and our condition on , we get that • Contradiction

  33. Theorem 10 • Given over , the decoding algorithm output in polynomial time the list of all codewords that differ from in at most: positions, where is the multiplicity parameter. The size of the list is at most , where: • Proof: bound on follows from lemma 6, bound on from bound on

  34. Parvaresh-Vardy General Case • Code is a subset of • Parameter changes: • Basis for over • defined by: • Encoding: , for , codeword is for • Bounds:

  35. That’s All Folks!

More Related