In information theory and coding theory, linear programming decoding (LP decoding) is a decoding method which uses concepts from linear programming (LP) theory to solve decoding problems. This approach was first used by Jon Feldman et al.[1] They showed how the LP can be used to decode block codes.
The basic idea behind LP decoding is to first represent the maximum likelihood decoding of a linear code as an integer linear program, and then relax the integrality constraints on the variables into linear inequalities.