In this paper, we extend the min-sum (MS) and its two improved algorithms (i.e., the normalized MS algorithm and the offset MS algorithm) to decode fountain codes over the binary input additive white Gaussian noise (BIAWGN) channel. We use Gaussian approximation method to analyze the asymptotic performance of fountain codes under various decoding algorithms and optimize the parameters of the two improved MS algorithms. Both the theoretical analysis and simulation results demonstrate that the normalized MS decoding with optimal parameter has better bit error performance than the offset MS decoding with optimal parameter.