The deep integration between non-orthogonal multiple access (NOMA) and massive multiple-input multiple-output (MIMO) technology is a promising way to achieve massive connectivity and high capacity in B5G/6G wireless communication system. In this letter, a novel signal processing framework for massive MIMO-NOMA data transmission is proposed based on group-level successive interference cancellation (GLSIC), in which the users are grouped according to the distances to the base station (BS) and NOMA is applied among different groups by employing GLSIC to reduce the inter-group interference. The uplink sum rate is derived for the proposed GLSIC-based massive MIMO-NOMA system with minimum mean square error channel estimator and maximal ratio combiner at BS, and the effects of channel estimation error, inter-group pilot contamination, and imperfect GLSIC are analytically quantified in the derivation. Theoretical analysis and simulation results show that, under the same conditions, the proposed GLSIC-based system achieves higher uplink rate than the existing cluster-based MIMO-NOMA system with user-level SIC.