The situation of mixed signals often occurs in the deteriorating electromagnetic environment, therefore how to separate the mixed signals successfully and efficiently has become a hot research in recent years. The high complexity of the per-survivor processing (PSP) algorithm that is used to separate the two phase shift keying mixed signals with the same frequency is studied. On the basis of previous studies, the single channel blind separation PSP algorithm of eliminating forward interference step-by-step for m-ary phase shift keying (MPSK) mixed signals is proposed. Simulation results show that, if the number of traversing states for every separation of a mixed signals is M^2(L-1) regarding as a measure of the complexity of the PSP algorithm, whereM is the modulation order of the signal, and L is the intersymbol interference length of two mixed signals, the complexity of the proposed algorithm for separating two binary phase shift keying (BPSK) mixed signals is four times lower than that of the original algorithm, and the complexity of separating two quadrature phase shift keying (QPSK) mixed signals is 16 times lower than that of the original algorithm. In the result, the proposed algorithm reduces the computational complexity of mixed signal separation greatly, so that the efficiency of separation of phase shift keying with the same frequency mixed signal has been improved.