Balancing Sampling Frequencies for Multi-modality IoT Systems: Smart Shoe as an Example

With the rapid development of Internet of Things, more and more sensing modalities are being integrated into single small devices. The data throughput, however, is always limited due to factors like transmission bandwidth and storage. How to balance the sampling frequencies among different sensing modalities, is little studied. Given a concrete system and its application, the question becomes: 1) how to qualitatively evaluate the loss of information as the sample frequency drops? 2) how to find the optimal sampling frequency ratio and set the multi-modality system according to it? In this paper, we will provide a general workflow to answer these questions. The workflow is evaluated on a self-developed smart-shoe, equipped with one pressure sensing matrix and two Inertial Measurement Units. This workflow could be interesting to both multi-modality IoT system developers and users, support them in finding the best frequency setting and obtaining datasets of higher qualities.