In this paper, we propose compressive sensing-based channel quantization feedback that is appropriate for practical massive multiple-input multiple-output (MIMO) systems. We assume that the base station (BS) has a compact (2-dimensional) uniform square array that has a highly correlated channel. To serve multiple users, the BS uses a zero-forcing precoder. Our proposed channel feedback algorithm can reduce feedback overhead as well as a codebook search complexity. Numerical simulations confirm our analytical results.