Abstract: An improved constrained total least squares approach was proposed for location of multiple targets. Firstly, the auxiliary variable is introduced to transform the nonlinear positioning equation into pseudo linear equation. Then the initial position of the target was estimated by two-stage weighted least squares, and the reference sensor was re-selected according to the initial estimation position. Finally, considering the noise of all the coefficients in the pseudo-linear equation and applying the Lagrangian multiplier to solve the constraint condition, an exact solution was obtained by using the quasi-Newton iterative formula. Simulations show that the proposed approach can reach the Cramér-Rao lower bound with strong robustness and precise positioning performance.

Key words: source location, time difference of arrival, frequency difference of arrival, Cramér-Rao lower bound