Abstract: Communication signals’ modulation recognition has great signality in non-cooperative communication. To investigate the automatic modulation recognition of high-order digital phase modulated signals widely used in satellite communication currently, such as QPSK, 8PSK ,16 APSK and 32 APSK , an algorithm for modulation recognition based on higher-order cumulants of signal and genetic algorithm is proposed. The paper recommends the different modulations in satellite communication, and gives the constellations of different modulations and the high order cumulants of different signals. It is shown that the higher-order cumulants which are higher than four-order can provide some useful information of signals through the recognition of QPSK, 8PSK and 16APSK, 32 APSK , which is based on higher-order cumulants . First, the modulation of QPSK, 8PSK and 16APSK, 32APSK will be recognized based on four and eight-order cumulants of signals. Second, the left signals will be treated with genetic algorithm, which is a fast-convergent and better stable genetic algorithm. It is the real number code and arithmetic crossover that are used in the genetic algorithm .The arithmetic crossover can effect the diversity in gene level in real number code under a certain condition, the ratio of signals’ amplitudes will be gained finally. Then the modulation of 16APSK and 32APSK could be recognized based on the arithmetical ratio of signals’ amplitudes with constellation characters, which is proved to be effective according to some simulation results. It is invariant with respect to difference of amplitude and phase, it is robust and widely usable. The efficiency of the algorithm is proved in theory, and the feature of the signal is more stable than other algorithms . Computer simulations show that the proposed recognition algorithm is effective in performance, and the application of this algorithm in low signal-to-noise ratio with a specified sample size results in high probability of correct identification(>96%).