A RECURSIVE ALGORITHM FOR TRACKING DOAS OF MULTIPLE MOVING TARGETS BYUSING LINEAR-APPROXIMATIONS

In this work, a new algorithm for tracking the directions-of-arrival ( DOA's) of moving targets by introducing a linear approximation is prop osed. The targets are assumed to move with constant angular velocities within a, short time and emitting continuously narrow-band signals th at impinge on an array of sensors. Therefore the trajectories of targe ts can be approximated by linear functions of lime, which consist of t he DOA's and the angular velocities, within the short time. In the con dition that the number of targets is known and the outputs vector of t he sensors including the additive white complex Gaussian noises is obs erved continuously, a cost function which consists of the squared resi dual error vectors is defined. The estimation of the DOA's and the ang ular velocities of targets is performed by minimizing this cost functi on. By estimating both the DOA's and the angular velocities at the sam e time, the proposed algorithm is able to improve the tracking perform ance for rapidly moving targets. In computer simulations, the performa nce of the proposed algorithm is compared with the ESPRIT method, whic h is one of the typical subspace methods with super resolution.