To investigate the occurrence mechanism of rockfall motion and accurately analyze the motion forms at each moment of the rockfall motion for the prevention and control of rockfall hazards, microelectromechanical inertial navigation detectors and three different signal decomposition algorithms including empirical mode decomposition (EMD), variational mode decomposition (VMD), and empirical Fourier decomposition (EFD) were employed. The study explored the acceleration waveform of a single motion state during the rockfall and the decomposition performance of the three signal decomposition algorithms for mixed motion acceleration signals. In the algorithm simulation tests, the empirical Fourier decomposition exhibited the smallest root mean square error compared to the other algorithms. In laboratory rockfall experiments, the sub-signals decomposed by the empirical Fourier decomposition showed the strongest interpretability, and the algorithmic complexity was significantly lower than that of the other algorithms. Therefore, the empirical Fourier decomposition can be used to further analyze the acceleration signals in microelectromechanical inertial navigation systems. The empirical Fourier decomposition is an effective and accurate algorithm for studying the motion state of rockfall.