Generation of Antipodal Random Vectors With Prescribed Non-Stationary 2-nd Order Statistics

IEEE Trans. Circuits Syst. I

Pareschi

Cambareri³

et al. 2017

Self Cite

Compressed Sensing (CS) can be introduced in the processing chain of a sensor node as a mean to globally reduce its operating cost, while maximizing the quality of the acquired signal. We exploit CS as a simple early-digital compression stage that performs a multiplication of the signal by a matrix. The operating costs (e.g., the consumed power) of such an encoding stage depend on the number of rows of the matrix, but also on the value and position of the rows' coefficients. Our novel design flow yields optimized sparse matrices with very few rows. It is a non-trivial extension of the rakeness-based approach to CS and yields an extremely lightweight stage implemented by a very small number of possibly signed sums with an excellent compression performance. By means of a general signal model we explore different corners of the design space and show that, for example, our method is capable of compressing the signal by a factor larger than 2.5 while not considering 30% of the original samples (so that they may not be acquired at all, leaving the analog front-end and ADC stages inactive) and by processing each of the considered samples with not more than three signed sums

Section: Rakeness-based Ternary Csmentioning

confidence: 99%

“…The generation of those rows, i.e., of ternary vectors with a prescribed correlation matrix, can be pursued by generalizing the methods proposed for antipodal vectors (see, e.g., [31], [32], [33], [28]).…”

Section: A Unstructured Zeroingmentioning

confidence: 99%

Rakeness-Based Design of Low-Complexity Compressed Sensing

IEEE Trans. Circuits Syst. I

Pareschi

Cambareri³

et al. 2017

Self Cite

“…In particular, by letting x P R n be a Random Vector (RV) that models a signal ensemble, we say that it is localised if its correlation matrix C x is non-white [8]. With this hypothesis, recent contributions [7], [8], [19] have shown how 2 This choice perfectly quantises the full range of the j-th measurement y j P r´Y, Y s, where Y " 2 bx´1 n " maxx ř n´1 l"0x l with }x}8 ď 2 bx´1 . This is the author's version of an article that has been published in this journal.…”

Section: B Set Partition Coding Of Wavelet Coefficientsmentioning

confidence: 99%

A Case Study in Low-Complexity ECG Signal Encoding: How Compressing is Compressed Sensing?

Cambareri

IEEE Signal Process. Lett.

Pareschi

et al. 2015

Self Cite

When transmission or storage costs are an issue, lossy data compression enters the processing chain of resourceconstrained sensor nodes. However, their limited computational power imposes the use of encoding strategies based on a small number of digital computations. In this case study, we propose the use of an embodiment of compressed sensing as a lossy digital signal compression, whose encoding stage only requires a number of fixed-point accumulations that is linear in the dimension of the encoded signal. We support this design with some evidence that for the task of compressing ECG signals, the simplicity of this scheme is well-balanced by its achieved code rates when its performances are compared against those of conventional signal compression techniques.

“…While the generation of a random sequence with an advised correlation is quite easy, the generation of antipodal sequences given its statistical characterization is a more complex task. In this field, different approaches have been proposed in literature [26], [27], [28], of which the simplest one relies on thresholding of Gaussian random vectors [26], [29]. Although not completely general, this approach is very simple and can be easily specialized to our case as in [21].…”

Section: Rakeness-based Csmentioning

confidence: 99%

Energy-Aware Bio-Signal Compressed Sensing Reconstruction on the WBSN-Gateway

Bortolotti

IEEE Trans. Emerg. Topics Comput.

Bartolini

et al. 2018

Self Cite

Technology scaling enables today the design of ultra-low power wearable bio-sensors for continuous vital signs monitoring or wellness applications. Such bio-sensing nodes are typically integrated in Wireless Body Sensor Network (WBSN) to acquire and process biomedical signals, e.g. Electrocardiogram (ECG), and transmit them to the WBSN gateway, e.g. smartphone, for online reconstruction or features extraction. Both bio-sensing node and gateway are battery powered devices, although they show very different autonomy requirements (weeks vs. days). The rakeness-based Compressed Sensing (CS) proved to outperform standard CS, achieving a higher compression for the same quality level, therefore reducing the transmission costs in the node. However, most of the research focus has been on the efficiency of the node, neglecting the energy cost of the CS decoder. In this work, we evaluate the energy cost and real-time reconstruction feasibility on the gateway, considering different signal reconstruction algorithms running on a heterogeneous mobile SoC based on the ARM big.LITTLE TM architecture. The experimental results show that it is not always possible to obtain the theoretical QoS under real-time constraints. Moreover, the standard CS does not satisfy real-time constraints, while the rakeness enables different QoS-energy trade-offs. Finally, we show that in the optimal setup (OMP, n = 128) heterogeneous architectures make the CS decoding task suitable for wearable devices oriented to long-term ECG monitoring.