During the last decade, Volterra Model has been considered as the most Usual and popular way to handle non linear systems. As ail alternative to this modelling strategy the last few years has registered the birth of a new modelling technique developed on a particular Hilbert pace the kernel of which is reproducing. This space known as Reproducing Kernel Hilbert Space (RKHS) uses the statistical learning theory to provide ail RKHS model. This paper proposes the modelling of a non linear Multi Input Multi Output (MIMO) communication channel with two modelling techniques reduced MIMO Volterra model and RKHS MIMO. the performances of both models are evaluated and the results were successful.