A series of large-scale direct shear tests (DST) was conducted on sand-contaminated ballast (SCB) in order to address the paucity of research on subgrade fillers contaminated with sand. Wireless attitude sensors, known as SmartRock, were used to monitor the internal particle movement during the shear process of the SCB. The results show that the shear strength of the SCB decreases with an increase in the void contamination index (VCI), leading to changes in the shape of the shear stress-shear strain curve of the SCB. Additionally, the power function offers a more accurate fit for the peak shear strength envelope diagram of the SCB at various VCI levels. The volumetric strain exhibits a pattern of shear compression-dilation, and the power function better predicts the critical shear strain and peak volumetric strain in the shear compression-dilation process of the SCB specimens. The SmartRock sensor can capture the motion pattern of individual particles within the shear plane. The DST accurately reflects the behavior of the real ballast bed under train loading conditions. Moreover, the study shows that the motion pattern of particles within the ballast layer is multidimensional. The interaction forces between particles in the ballast bed, perpendicular to the train travel direction, can be described by unidirectional shear action.