Scatter contamination in cone-beam computed tomography (CBCT) degrades the image quality by introducing shading artifacts. In our previous study, a moving-blocker-based approach was proposed to simultaneously estimate scatter and reconstruct the complete volume within field of view (FOV) from a single CBCT scan. Promising results were obtained from simulation studies. In this work, we implemented the moving blocker system on a LINAC on-board kV CBCT imaging system. A physical attenuator (i.e., "blocker") consisting equal spaced lead strips was mounted on a linear actuator. A step motor connected to the actuator drove the blocker to move back and forth along gantry rotation axis during CBCT acquisition. Scatter signal was estimated from the blocked region of imaging panel, and interpolated into the un-blocked region. A sparseness prior based statistical iterative reconstruction algorithm was used to reconstruct CBCT images from un-blocked projections after the scatter signal was subtracted. Experimental studies were performed on both a Catphan phantom and an anthropomorphic pelvis phantom to evaluate performance of the moving blocker system. The scatter-induced shading artifacts were substantially reduced in the images acquired with the moving blocker system. CT number error reduced in selected regions of interest of the Catphan phantom from 318 to 17. It also decreased in those of the pelvis phantom from 239 to 10. We demonstrated for the first time that the moving blocker system could successfully estimate the scatter signal in projection data, reduce the imaging dose and obtain complete volumetric information within the FOV using a single scan.