On channel estimation in cell-free massive MIMO for spatially correlated channels with correlated shadowing under Rician fading

This work addresses channel estimation (CE) in the uplink phase for a cell-free massive multiple-input multiple-output system operating under the time division duplex protocol. Considering that, channels are spatially correlated under the Rician fading model, where the investigated model is composed of two components: deterministic and nondeterministic, with the deterministic component describing the line-of-sight paths and the nondeterministic component describing the non-line-of-sight paths. Additionally, we dealt with correlated shadow fading that represents the most realistic situation. On the other hand, this work introduces a dynamic cooperation cluster framework in which the user is not served with the whole network ( i.e., all access points [APs]) but only the APs that present the best channel conditions regarding that user. In other words, this work proposes partial CE for each user because only APs with the best channel conditions are allowed to compute channel estimates. Consequently, we proposed partial channel estimators that perform the CE process with low complexity, namely, a partial minimum mean square error estimator and a partial element-wise minimum mean square error estimator. In addition, a simple pilot assignment technique is proposed in order to reduce interference signals so that each user experiences low interference from other users. Furthermore, the computational complexity required by each estimator is derived, where it is represented by the number of complex multiplications that each estimator requires in each consistency block. Theoretical and simulated results are provided, where the performance of each estimator is evaluated and analyzed using the normalized mean-square error metric.