The ADIS16367 has an analogue bandwidth of some 330[Hz] with the possibility to turn on low-pass filters. This can be done for the OpenShoe units with a provided Matlab control script. Of cause, higher bandwidth means more noise and there is no reason to have a higher bandwidth than necessary. Therefore, I thought I would post some spectra of the inertial measurements taken from normal walking. Below, the spectra of the specific force along the x-axis of the IMU is seen:
The frequency content along the other axis are similar as well as for the angular rate. The conclusion is that most of the power if located below 200[Hz] with the great bulk below 30[Hz]. This means that, for walking motions, low-pass filters can preferably be turned on (or sensors with lower bandwidth can be used). Below is a zip-archive for reference with the remaining spectra in form of Matlab figures: