The traffic matrix is one of the crucial inputs in many network planning and traffic engineering tasks,it is agreed that accurate traffic matrices are crucial,but it is usually impossible to directly measure traffic matrices. So,it is an important research topic to infer traffic matrix by reasonably modeling,and incorporating the measurement data of limited links,as well as other prior information.This paper presents the basic concept of traffic matrix and its estimation mechanism,categorizes and analyzes more than twenty different approaches to traffic matrix estimation problem proposed recently,and summarizes current research achievements on traffic matrix estimation problem.This paper also discusses the performance and estimation errors of current approaches. Finally,future research directions and potential applications of these researches are identified.