Enabling High-Speed Data Acquisition with Compressive Sampling

We augmented an existing inertial measurement unit, IMU, with a secondary module sampling at high speeds (100Hz) an array of 1260 force sensitive resistors, FSR, embedded in a flexible, thin plastic foil. Due to legacy imposed by the existing sensor platform, we were confronted with severe bandwidth restrictions on the IC channel between the two modules. The secondary module created data at about 1.2MBit/s, however the IC component of the IMU had an upper limit of 400kBit/s. Communication between the two modules was necessary as storage and wireless communication were only available on the IMU module. Since altering the existing system was not an option, we looked for ways to reduce the bandwidth requirements while at the same time maintain the core functionality of the secondary module. We present in this work our analysis of a Compressive Sampling (CS) solution for our specific sensor setting. We tested variations of CS and present their performances. To increase sparsity in the input data, we applied a differential encoding scheme on the data frames. The original CS algorithm was applied to the difference frames without further pre-transformation. We also tested the performance of a Wavelet-transformation step. Finally, we implemented on the embedded sensor platform a differential encoding scheme without pre-transformation and thus reduced the bandwidth requirements to about 30% of the original demands.