Abstract: In order to better estimate the instantaneous frequency of signals, the empirical mode decomposition (EMD) algorithm, proposed by Huang et al., is used to break multi-component signals into several narrow subbands. EMD is an adaptive method and can be used to analyze nonstationary signals, so it has been widely applied to many engineering fields. However, EMD is still considered as an empirical method because it lacks a rigorous mathematical foundation, and its analysis depends largely on numerical simulations and experimental investigations. In this paper, related problems of the EMD algorithm are discussed, including its theoretical foundation and its applications. Some modified EMD algorithms are considered to overcome problems, such as stopping criterion, end effect, envelope of signals and mode aliasing. The applications of EMD to the processing of radar signals are reviewed. Some directions for further research on the EMD algorithm are suggested.