The paper presents the results of work on reducing the motion data dimension obtained by motion capture system with inertial microelectromechanical sensors. In the beginning, it is given which movements were chosen and rationale for everyone's. After that, assumption is presented about independence of data flows for different types of sensors is checked. Next, method for sorting the incoming data stream for the wavelet transform is considered and method for redundancy decline through finding strong linear relations between data streams is given and scale of wavelet transform is justified. Then method for compressing motion data using a wavelet transform is proposed. After that, it is shown that after data processing, data power spectrum decreases in the high-frequency region.