Adaptive user grouping algorithm for the downlink massive MIMO systems

SINR loss and user selection in massive MU‐MISO systems with ZFBF

“…Three methods reducing the loss caused by SC were compared with the method proposed in Section 5 (SSCV). The first was the random method, which randomly selects users from the user set [12]. The second method we investigated was that which eliminates the maximum correlation value (EMCV), which can be regarded as the semi-orthogonal method.…”

“…Because the method using the ACV with a group has the same results as with SCV, we show only the results of SCV here. Note that the ideas of EMCV [9][10][11], ACV [13][14][15], and SCV [12,16] are all widely used in studies to reduce the performance loss caused by SC in a group. Figure 6A and 6B show the SINR loss in the i.i.d.…”

“…To avoid performance loss, only semi-orthogonal users can be served in the same group. In addition, many studies use metrics to estimate the loss caused by SC [12][13][14][15][16]. In these studies, the average of correlation values (ACV) or sum of correlation values (SCV) between users Separating highly correlated users can reduce the loss caused by spatial correlation (SC) in multiuser multiple-input multiple-output (MU-MIMO) systems.…”

“…In other words, users in the same group should have small correlation values to reduce their losses caused by SC. Note that these methods based on CoC are also usable in massive MIMO systems because scenarios exist in which some users are highly correlated [12,17,18].…”

“…In previous works, although CoC has been widely used to reduce the loss, practically no research exists on the accurate loss caused by CoC, particularly in cases with more than two users. Furthermore, most studies use the average or sum of CoC to estimate the effect of correlation on user performance [12][13][14][15][16]. For example, the authors in [16] consider that the sum of CoC should be calculated to estimate the SC effect, but they do not show the reason for it.…”