A 232-1996-kS/s Robust Compressive Sensing Reconstruction Engine for Real-Time Physiological Signals Monitoring

• Published in: IEEE journal of solid-state circuits 2019-01, Vol.54 (1), p.307-317
• Main Authors: Chen, Ting-Sheng, Kuo, Hung-Chi, Wu, An-Yeu
• Format: Article
• Language: English
• Subjects: Biomedical signal processing; Biosensors; CMOS; compressed sensing; Engines; Integrated circuits; Length measurement; multiple atom search; Noise levels; Noise measurement; Noise sensitivity; Physiology; Real time; Real-time systems; reconfigurable architecture; Reconstruction algorithms; Remote sensors; Robustness; Semiconductor device measurement; Signal monitoring; Signal reconstruction; Time compression; Upgrading; Very large scale integration; Wireless sensor networks
• ISSN: 0018-9200; 1558-173X
• DOI: 10.1109/JSSC.2018.2869887

Compressive sensing (CS) techniques enable new reduced-complexity designs for sensor nodes and help reduce overall transmission power in wireless sensor network. However, for real-time physiological signals monitoring, the orthogonal matching pursuit that applied prior CS reconstruction chip designs is sensitive to measurement noise and suffers from a low convergence rate. In this paper, we present a robust 232-1996-kS/s CS reconstruction engine fabricated in 40-nm CMOS. With combination sparsity estimation (SE) and robust subspace pursuit (SP), more than 8-dB signal-to-noise ratio (SNR) gain is achieved under the same success rate for robust reconstruction. For hardware implementation, a flexible indices-updating VLSI architecture inspired by the gradient descent method can support arbitrary signal dimension of CS reconstruction without matrix decomposition. Parallel searching, indices bypassing, and processing elements (PEs) grouping are designed to reduce the total CS reconstruction cycle latency, thus enhancing the throughput rate by approximately 6.3 for CS reconstruction. The 8.66-mm CS reconstruction engine can provide real-time physiological signal reconstruction for data collected from CS-based wireless biosensors under noisy conditions, making low-power patient monitoring a reality.