: Due to platform motion in airborne radars, relative velocity of radar and ground is not zero, result in non-zero doppler shift of ground clutter. In the side-looking airborne phased array radars, there is a linear relation between normalized angle of arrival and normalized doppler frequency of ground clutter patchs. On the other hand, in GMTI mode ground clutter arrives from sidelobe region of the beampattern in wide range of angles. So clutter spectrum will be two dimensional and extends along a narrow ridge in angle-doppler plane. Therefore, conventional methods like beamforming and doppler processing are not efficient for clutter rejection, andjoint spatial and temporal processing methods like STAP are required. Clutter spectrum is basically a narrow ridge, therefore the narrow notch created by STAP filter can effectively mitigate the clutter and permit detection of small and slow targets that might otherwise be obscured by sidelobe clutter. STAP methods are devided into two kinds, including statistical and non-statistical. In this thesis after a review on statistical STAP methods, one of the most important non-statistical STAP approach called direct data domain (D 3 ) and its variants are described. It has the advantages of no secondary data and interference estimation requierment and the ability of discrete interference mitigation. Also, some disadvantagesare reduced degrees of freedom (DOF), weak homogenous interference mitigation capability and high sidelobe levels. In this thesis, a novel methodbased on a convex optimization framework isproposedtoreduce sidelobes of D 3 approach with integrated sidelobe level. Simulation results show better performance of proposed method than that forconventional D 3 method in sidelobe reduction of beampattern in all of ISL, PSL and half power beamwidth criteria. The other advantage of the proposed method is that it has deeper beampattern nullsin the direction of interferences. Also due to the unwanted shift of D 3 beampattern using CG method in low SNR values, a new method called LSMR is used to find adaptive weights in forward D 3 processor. Keywords: Airborne radar, Ground clutter, Space-Time Adaptive Processing , Direct data domain
