The invention discloses a large-scale MIMO capacity improving method based on unmanned aerial vehicle cluster deployment, and the method comprises the steps: firstly enabling each single-antenna unmanned aerial vehicle to be randomly deployed in an area above a multi-antenna ground base station, and enabling each unmanned aerial vehicle to assist in estimating channel state information through a geographic position system; randomly selecting one unmanned aerial vehicle to communicate with the neighbor unmanned aerial vehicle, constructing local information, and calculating the current income; learning deployment behaviors according to earnings, and keeping the positions of other unmanned aerial vehicles unchanged; and finally, determining the optimal deployment position of each unmanned aerial vehicle after a plurality of rounds of interaction. According to the invention, the channel capacity is improved by adjusting the deployment position of each unmanned aerial vehicle; and deployment is completed only by using local information without an optimization control center, so that the communication energy consumption is reduced, the communication control time delay is shortened, and the endurance and real-time control capability of the unmanned aerial vehicle cluster are improved. Applicable wireless communication scenarios include, but are not limited to, high density venue communications, battlefield communications, disaster relief and rescue communications.