Heartbeat-like noises in the speed estimations of cf's

[girayus]

Hi,
In my work, I use a central computer to calculate velocity inputs of cf's and send them using cmdVelocityWorld command. I'm using Vicon with single motion capture markers for the position data. For the motion planning algorithm I use, speeds of the cf's are required.

By following the suggestions in the previous discussions, I used position data to calculate the velocities of the cf's using discrete differentiation, i.e. v[k] = (p[k]-p[k-1])/dt where k is the time step and dt=1/sleepRate. Even though the resulting speed estimations are pretty accurate most of the time, heartbeat-like spikes occur at some time steps.
I ran a simple test by sending constant velocity commands with 0.5m/s speed to a single cf for 200 time steps at sleepRate=20 Hz. Vicon was running at 120Hz. There was no latency warning after the take-off. These are the speed (calculated using discrete differentiation) vs. time step plot, terminal logs, and the test script:

terminalogs.txt
python_test_script.txt

When I tested it with multiple cf's, I observed that these spikes occur at the same time step for all cf's. What could be the reason for these heartbeat-like noises? Any suggestions to avoid them?