Abstract: Ultrasonic guided waves are very sensitive to the material's characteristics and the change of the cortical thickness in long bones.Therefore,it has been a great interest of using guided waves to assess long bones.However,it's a challenge to acquire the group velocity and other parameters from the long bone because of the mixed multiple modes in the received signals.In the study,the spatial time-frequency based blind source separation(TFBSS)method was employed to separate the mixed signals and then calculate the group velocities of each single mode.The result showed that the TFBSS method can separate the mixed signals correctly,and the experimental measured group velocities are agreed with theory values very well.The cortical thickness could be acquired based on the group velocities,which is quite significant to assess the properties of long bones.