Multi-user multiple-input and multiple-output (MU-MIMO) transmission scheme provides potential gain in terms of system capacity in spatial correlated scenarios and is considered as one of the key technologies in Long Term Evolution (LTE) systems. The goal of this paper is to gain insight into MU-MIMO detection concepts and their implementation aspects. Various receiver architectures are investigated and their performance are assessed through an advanced link-level simulator. The motivation to study the MU-MIMO detectors rises from the weakness of the co-channel interference-unaware single user detector, the conventional detector for single user MIMO transmission. The most essential problem of these conventional single user detectors in MU-MIMO is the failing to detect signals under the residual interference caused by co-scheduled user, which reduces the performance of the detector. Different practical scenarios have been considered in the evaluation process, namely low and high spatial correlation channels with real channel estimation and feedback delay. Our investigation has shown that single user receivers, such as maximum ratio combiner, show generally poor performance in MU-MIMO scenario and as such they should be neglected for practical implementation. Furthermore, we have shown that the interference rejection combiner achieves the best performance-complexity tradeoff and outperforms conventional single user detectors only with a small increase of computation efforts in signal processing.