Fault-tolerance has been widely studied these years in order to fit new kinds of applications running on unreliable systems such as the Internet. Erasure coding aims at recovering information that has been lost during a transmission (e.g. congestion). Considered as the alternative to the Automatic Repeat-reQuest (ARQ) strategy, erasure coding differs by adding redundancy to recover lost information without the need to retransmit data. In this paper we propose a new approach using the Finite Radon Transform (FRT). The FRT is an exact and discrete transformation that relies on simple additions to obtain a set of projections. The proposed erasure code is Maximal Distance Separable (MDS). We detail in this paper the systematic and nonsystematic implementation. As an optimization, we use the same algorithm called "row-solving" for creating the redundancy and for recovering missing data.