Title:Gabidulin Decoding via Minimal Bases of Linearized Polynomial Modules

Abstract:We show how Gabidulin codes can be decoded by using a parametrization approach. Our decoding algorithm computes a list of all closest codewords to a given received word. We consider a certain module, called the interpolation module, over the ring of linearized polynomials with respect to composition of polynomials. We consider minimal bases of this interpolation module and show that the Predictable Leading Monomial (PLM) property holds in this setting. We show that any such minimal basis gives rise to a parametrization, using the PLM property as a key ingredient in our proof. Two subalgorithms are presented to compute such a basis, one iterative, the other an extended Euclidean algorithm. Both of these subalgorithms have polynomial time complexity. The complexity of the overall algorithm, using the parametrization, becomes exponential as soon as the closest codewords are beyond the unique decoding radius. However, for high rate codes, this complexity beats the complexity of exhaustive search.