Lattice based cryptography is claimed as a serious candidate for post quantum cryptography, it recently became an essential tool of modern cryptography. Nevertheless, if lattice based cryptography has made theoretical progresses, its chances to be adopted in practice are still low due to the cost of the computation. If some approaches like RSA and ECC have been strongly optimized -in particular their core arithmetic operations, the modular multiplication and/or the modular exponentiation -lattice based cryptography has not been arithmetically improved. This paper proposes to fill the gap with a new approach using Residue Number Systems, RNS, for one of the core arithmetic operation of lattice based cryptography: namely solving the Closest Vector Problem (CVP).