Adaptive Compressive Sensing Techniques for Low Power Sensors | Consortium for Embedded Systems | SIU

Abstract: The objective of this project is to investigate novel adaptive approaches to more effectively apply compressive sensing (CS) techniques [1, 2] in low-power sensor systems. Recently, CS emerged as an attractive technique in low-power sensor development [3, 4, 5], because of its capability to allow sensor signals to be sampled at rates lowers than Nyquist rate. In CS operation, the number (denoted as M) of data points to be sensed or transmitted is selected according to the sparsity of the sensor signals. Currently, the majority of CS sensor circuits assume fixed M values during the entire sensing operations. To further reduce the power consumption of CS sensors, we have been investigating the potentials of adaptively adjusting the sampling rates in CS operations at system level. As a natural extension to the current effort, the proposed research is to investigate circuit techniques to execute adaptive compressive sensing at sensor nodes. Particularly, we will investigate the use of low-power analog wavelet transformation circuit to detect when sampling rate can be changed in adaptive CS operations.

Problem: In CS operation, the measurement size (or sampling rate) is affected by signal sparsity, which may vary over the time. Hence, it is naturally expected that the CS measurement size (or sampling rate), should follow the variations of signal sparsity. Such CS operation is called adaptive CS (ACS). At present, the sampling rates of the reported CS sensors are fixed according to prior knowledge of signal sparsity, and hence do not change during sensing operations. By exploiting signal sparsity variations, ACS can potentially lead to further power reduction in low-power sensors.

Rationale / Approach: The findings in the early phase of the project indicate that there are significant signal sparsity variation over the time and applying ACS potentially lead to significant power saving. The proposed work intends to develop techniques to detect the signal sparsity variations and subsequently adjust the sampling rate. We propose to use a low-power analog (continuous-time) wavelet transformation (WT) circuit to monitor the signal sparsity during sensor operation and to adjust the sampling rates according to the output of the analog WT circuit.

Novelty: CS is a relatively new approach to reduce the power consumption of certain type sensors. The proposed work helps make CS sensors more power efficient, hence advances the state of the art of CS sensors.

Potential Member Company Benefits: The developed techniques have the potentials to help member companies further reduce the power consumptions of certain sensor devices in productions or to be used in their research and development (R&D) projects.

Deliverables for the proposed year:
1. Design of the analog wavelet transformation circuit
2. Investigation results on the effectiveness of using analog wavelet transformation circuit to determine the sampling rates in adaptive compressive sensing

Milestones for the proposed year:
1. Quarter 1 (08/14-10/14): Select a set of wavelet functions to be implemented
2. Quarter 2 (11/14-01/15): Obtain stable transfer function and start analog WT circuit development work
3. Quarter 3 (02/15-04/15): Complete the development of analog WT circuits
4. Quarter 4 (05/15-07/15): Establish the relation between analog WT circuit output and signal sparsity