Abstract: In view of the time-difference-of-arrival sound-source direction finding problem,if the velocity parameter of propagating signal about sound source is known,combining with estimated result of the Direction of Arrival (DOA) of Linear Least Squares (LLS),the direction finding problem can be solved by iterative algorithm based on maximum likelihood principle.However,in some conditions,it is difficult to obtain velocity parameter of sound source,or measured deviation is large,which will affect the estimation accuracy of DOA.According to the characteristic of LLS about direction finding without sound velocity,and the characteristic of iterative algorithm based on maximum likelihood principle directly solving the azimuth and elevation of sound source,this paper proposed an iterative model without sound speed.The Taylor Series expansion method and the Levenberg-Marquart method are used to solve the problem,without knowing the velocity parameters of the sound signal beforehand.In addition,the mean square error of direction finding about Taylor Series expansion method and the Levenberg-Marquart method,and the Cramer-Rao Lower Bound of iterative model without sound speed are derived.The simulation results show that when measurement errors exist in sound speed,the proposed method is obviously superior to the existing methods.

Key words: Cramer-Rao lower bound, Direction finding, Iterative model without sound speed, TDOA