Abstract: ‍ ‍In order to solve the problem of limited network resources and the difficulty of effective resources allocation in large-scale unmanned aerial vehicle （UAV） network， a three-dimensional （time domain， frequency domain， and spatial domain） network resource allocation algorithm based on graph coloring was proposed， which can cope with arbitrary number of UAV transceiver pairs. Specifically， the retro-directive array antenna was adopted to utilize the spatial dimension resource on the basis of the traditional time-frequency network resource partition， which leaded to a three-dimensional network resource partition problem. The problem was then modeled to a graph coloring problem. Heuristic graph coloring algorithm， greedy graph coloring algorithm， and the corresponding mapping algorithm from coloring results to network resource allocation were proposed. Simulation results show that the proposed algorithm greatly improves the throughput and packet success rate compared with the traditional time-division multiple access （TDMA） and time-frequency resource allocation.