Abstract: In order to reduce the bit error rate of multi access communication systems, the multi-target spot resolution technology based on four quadrant detectors was used to analyze the characteristics of communication laser spot data streams, and three traditional clustering algorithms were compared. The k-means clustering algorithm, which exhibited better comprehensive performance in multi spot resolution, was extended to propose a drift based multi spot clustering algorithm based on data streams. The optimal number of clusters was selected through initial clustering first adaptively, then real-time drift detection and clustering were performed on new spot data, and the classification decision parameters of the algorithm were updated in real time. The results show that, the problem of multi spot resolution under spot drift is solved by this algorithm, and the spot resolution accuracy is significantly improved compared to traditional algorithms, stabilizing at over 90%. This study has improved communication quality and provided algorithm support for the implementation of multi access communication.