Aiming at the problem that using Baum-Welch（ BW） algorithm in the hidden Markov model（ HMM）algorithm is difficult to find the global optimal solution when predicting the model parameters

a particle swarm optimization-hidden Markov model（ PSO-HMM） algorithm based on particle filtering and hidden Markov is proposed. PSO is added for the initial value selection problem in BW algorithm learning. Optimization process of PSO

the quantization of the initial value selection for the channel state information（ CSI） wireless signal positioning data processing is more reasonable and thus increases the accuracy of positioning. At the same time

in order to overcome the statistical constraint problem of HMM algorithm

a renormalization and remapping mechanism is added. Since the convergence speed of PSO is faster

from the macroscopic phenomenon aspect

the real-time positioning is improved. The experimental results show that compared with the traditional HMM algorithm and support vector machine（ SVM） algorithm

the PSO-HMM algorithm greatly improves the accuracy and real-time performance of indoor positioning.