Based on the characteristic that the Bistatic forward scattering radars are superior to the monostatic radars in effectively detecting and tracking the small radar cross section (RCS) targets, non-variable acceleration and variable acceleration motion models are constructed. The Extended Kalman Filter algorithm is applied to keeping targets in track, the parameters (trace of the target, velocity and acceleration of the targets) of the moving targets are accurately estimated which establish the foundation for radar imaging and identification. By applying Gauss-Newton iterative algorithm, the efficiency and accuracy of the filtering results are improved. The models proposed and the effectiveness of the algorithms are verified by simulating the non-variable and variable acceleration motion targets.