Velocity change during step cycle

I have tested one manufactures laser radar. The subjects in my tests have been runners and long jumpers. The sampling frequency has been from 500 to 750 Hz. The raw signal is quite noisy, but after filtering I’m able to find sini wave like velocity curve during each step cycle and detect differences between the legs.
Now I try to optimize the filtering method. Does any of you know, or have references, how much the runners’ velocity changes during one step cycle (flight/contact phase) at the running speeds of 8 – 11 m/s?