In massive multiple-input multiple-output (MIMO) systems, user clustering techniques are crucial for addressing inter-beam interference. These techniques become even more critical when incorporating non-orthogonal multiple access (NOMA) to enhance spectral efficiency and user fairness. Prior research in this area has primarily focused on channel correlation for user pairing. This paper presents a novel approach for user pairing and subband allocation in a crowded downlink system that utilizes a criterion based on minimizing the condition number (CN) of channel matrices. A benchmark resource allocation technique, which is based on a rate maximization criterion and has a higher complexity but quasi-optimal performance, is also introduced. To evaluate the performance of the proposed technique in a realistic setting, experimental massive MIMO channel measurements in a dense user deployment scenario are used. Results indicate that the CN criterion leads to significantly higher throughput and fairness levels compared to the channel correlation criterion and achieves performance that is close to that obtained with rate maximization. The study is further extended to the case of multi-antenna reception, where interference cancellation techniques are proposed to increase the throughput of users with low channel gains. The proposed methods outperform a well-referenced technique in the literature.