Network tomography is a novel technique of network measurement in recent years. It combines network measurement with statistical inference, which can solve some problems that can not be solved by normal network measurement techniques. Network topology inference is one of the main applications of network tomography, which can infer the topology of network through end to end measurement and does not need the cooperation of internal nodes. In this paper, the technique of topology inference based on network tomography is summarized systematically, and the topology inference techniques and algorithms are compared and analyzed. Finally, the problems of current research and the future directions are discussed.