Abstract: In order to research the power spectrum characteristics of M-ary Position Phase Shift Keying (MPPSK) modulation signal deeply, and improve the demodulation performance of the receiver, the theoretical power spectrum expression of MPPSK is presented when M is larger than 2, and a new MPPSK demodulator based on BP neural network is designed. Based on the principle of MPPSK modulation, the decomposition method of alternating wave and steady wave is utilized to deduce the theoretical power spectrum expression of MPPSK when M equals 3, and the expression is extended to the MPPSK when M is larger than 3. The simulation is conducted to verify the correctness of theoretical derivation, and the main factors impacted the power spectrum of MPPSK is also analyzed. Meanwhile, based on the two existing demodulation methods of phase-locked loop (PLL) demodulation and mutli-path discrimination relying on the digital impacting filter, the MPPSK demodulator based on BP neural network is presented by utilizing the obvious difference of the output of digital impacting filter, and the corresponding simulation are also carried out. Simulation results show that the new MPPSK demodulator has a better demodulation performance than the two existing demodulation schemes.