Incremental Random Sampling Based Clustering and Prediction Algorithm for User Distribution Analysis in NR Coverage Optimization System

Currently, the commonly used coverage optimization system based on Massive MIMO for NR network generally uses the delayed user distribution data, lack of certain predictability, and the large scale of user distribution data leads to slow analysis. In order to solve the problem that the traditional density-based clustering algorithm is difficult to meet the requirements of the frequently re-clustering in the scenario of high frequency rolling update of 5G MR Data, which leads to low analysis efficiency, the concept of incremental random sampling (IRS) is proposed, and a kind of density-based clustering algorithm based on incremental random sampling (IRS-DBSCAN) and a user distribution prediction algorithm based on incremental random sampling (IRS-UDPA) are proposed. The simulation and actual network verification results show that, compared with the traditional density clustering algorithm, the proposed algorithm has certain advantages, and it is feasible and practical to promote the actual network performance.