Pilot Allocation Algorithm Based on K-means Clustering in Cell-Free Massive MIMO Systems

In this letter, a pilot allocation algorithm based on K-means clustering is proposed for the cell-free massive MIMO systems. Firstly, to avoid the communication between users and “invalid” access points (APs), a user-centered virtual cell division is adopted. Users select AP according to the large-scale factor of the channel to form a specific AP set which serves the specific users. Then, K value is determined by combining multiple features and the initial centroid is selected to cluster users. Finally, based on the results of user clustering, users in the internal cluster are assigned to orthogonal pilots, and users in different clusters are reused pilots, which can avoid pilot contamination (PC) caused by pilot reusing among users from the space perspective. Simulation results show that the proposed algorithm can reduce channel estimation error and enhance system throughput.