Abstract: This paper shows the variation of clusters with the increasing number of antennas. The data was collected from the massive MIMO mobile measurement at 3.5 GHz, in line of sight (LoS) and non line of sight (NLoS) conditions, respectively. And the virtual measurement method is used to form the 64-element, 128-element and 256-element virtual antenna array from the 32-element antenna array. After estimating parameters by the space-alternating generalized expectation maximization (SAGE) algorithm and clustering by KPowerMeans algorithm, the parameters of clusters are displayed in angular domain and delay domain. The cluster-level angular power spectrums (APS) are shown, the intra-cluster angular spread (AS) and intra-cluster delay spread (DS) of these 4 groups of data are calculated, to display the clusters’ variation when antenna number increases.