Discrete Mathematics
Algebraic Coding Theory: BCH and Reed-Solomon
A scratched CD still plays. A QR code with 30% obscured still scans. That is not magic - it is algebraic coding theory.
- **CD/DVD**: RS(255,239) over GF(2^8) corrects burst errors up to 4000 bits long, making surface scratches harmless
- **QR codes**: error-correction level H tolerates up to 30% damage thanks to Reed-Solomon over GF(2^8)
- **Flash memory**: BCH codes in NAND flash reduce unreadable-block rates from 10^-4 to below 10^-9
- **Deep space**: NASA used RS codes on Voyager and Cassini to correct errors across billions of kilometers
Предварительные знания
- Finite fields GF(q): arithmetic, irreducible polynomials, primitive elements
- Linear codes: matrix representation, Hamming distance, Singleton bound
- Polynomial rings and division with remainder
Finite Fields and Polynomial Codes
In 1960, Irving Reed and Gustave Solomon published the code that now protects every CD, DVD, and QR standard: at error densities up to 30%, the disc still reads correctly. The RS(255,239) code used in compact disc audio corrects up to 8 symbol errors per 255-symbol block.
How many errors does RS(255, 239) over GF(2^8) correct?
BCH Codes: Construction and Decoding
In 1960, Irving Reed and Gustave Solomon published the code that now protects every CD, DVD, and QR standard: at error densities up to 30%, the disc still reads correctly. The RS(255,239) code used in compact disc audio corrects up to 8 symbol errors per 255-symbol block.
What does Berlekamp-Massey find when decoding BCH/RS?
Algebraic Coding in Discrete Mathematics
Algebraic codes unify field theory, polynomial algebra, and linear spaces to solve the engineering problem of reliable data transmission.
- Group and field theory — GF(2^m) is a cyclic multiplicative group underpinning all code arithmetic
- Linear algebra — Codewords are vectors in a linear space over a finite field
- Complexity theory — Berlekamp-Massey runs in O(t^2) - polynomial-time decoding
- Information theory — RS codes are MDS codes achieving the Singleton bound: maximum correction capability at minimum redundancy
Итоги
- Galois field GF(2^m) is the foundation: 2^m elements, arithmetic modulo an irreducible polynomial of degree m
- RS(n,k) over GF(2^m): length n, k information symbols, minimum distance d = n-k+1, corrects t = floor((d-1)/2) errors
- BCH codes are built via LCM of minimal polynomials; generator polynomial roots include 2t consecutive powers of alpha
- Berlekamp-Massey finds the error locator polynomial from the syndrome sequence in O(t^2) field operations
- Chien search finds error positions as roots of the locator polynomial; Forney formulas give error values