The random frequency signal format not only reduces the instantaneous band-width and data sampling rate, but also has the good characteristics of the radio-frequency immunity. However, the frequency of the signal changes randomly, which makes the traditional methods of scattering-center estimation invalid. Herein, the model of random frequency signal echoes was derived, and a novel parameter estimating method for scattering-center model was proposed based on the analysis of the radar echoes′ sparse characteristic. The parameters estimation was converted to the problem of sparse signal reconstruction in the framework of Compressed Sensing(CS). The performance of the sensing matrix was investigated. Experiment results show that the presented method is effect even for low sampling-rate condition.