Gridding algorithm from non-uniform to cartesian sampling pattern has been extensively investigated for many years within MR imaging community. Recently, using regularization and sparsity constraint, a generalized version of gridding algorithm called NIRVANA(Non-iterative Regularized reconstruction Algorithm for Non-CartesiAn MRI) has been proposed. However, the NIRVANA has been optimized as a gridding method for single coil, and its performance is not optimal for parallel imaging applications. The main contribution of this paper is to extend the NIRVANA such that it is optimally tuned for coil-diveristy information, which results in coil specific gridding scheme. Extended experiments confirmed that our method outperforms the conventional ones due to coil specific density compensation matrices.