TY - GEN
T1 - A configurable 12-to-237KS/s 12.8mW sparse-approximation engine for mobile ExG data aggregation
AU - Ren, Fengbo
AU - Markovic, Dejan
N1 - Publisher Copyright:
© 2015 IEEE.
PY - 2015/3/17
Y1 - 2015/3/17
N2 - Compressive sensing (CS) is a promising solution for low-power on-body sensors for 24/7 wireless health monitoring [1]. In such an application, a mobile data aggregator performing real-time signal reconstruction is desired for timely prediction and proactive prevention. However, CS reconstruction requires solving a sparse approximation (SA) problem. Its high computational complexity makes software solvers, consuming 2-50W on CPUs, very energy inefficient for real-time processing. This paper presents a configurable SA engine in a 40nm CMOS technology for energy-efficient mobile data aggregation from compressively sampled biomedicai signals. Using configurable architecture, a 100% utilization of computing resources is achieved. An efficient data-shuffling scheme is implemented to reduce memory leakage by 40%. At the minimum-energy point (MEP), the SA engine achieves a real-time throughput for reconstructing 61-to-237 channels of biomedicai signals simultaneously with <1% of a mobile device's 2W power budget, which is 76-350× more energy-efficient than prior hardware designs.
AB - Compressive sensing (CS) is a promising solution for low-power on-body sensors for 24/7 wireless health monitoring [1]. In such an application, a mobile data aggregator performing real-time signal reconstruction is desired for timely prediction and proactive prevention. However, CS reconstruction requires solving a sparse approximation (SA) problem. Its high computational complexity makes software solvers, consuming 2-50W on CPUs, very energy inefficient for real-time processing. This paper presents a configurable SA engine in a 40nm CMOS technology for energy-efficient mobile data aggregation from compressively sampled biomedicai signals. Using configurable architecture, a 100% utilization of computing resources is achieved. An efficient data-shuffling scheme is implemented to reduce memory leakage by 40%. At the minimum-energy point (MEP), the SA engine achieves a real-time throughput for reconstructing 61-to-237 channels of biomedicai signals simultaneously with <1% of a mobile device's 2W power budget, which is 76-350× more energy-efficient than prior hardware designs.
UR - http://www.scopus.com/inward/record.url?scp=84940782753&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84940782753&partnerID=8YFLogxK
U2 - 10.1109/ISSCC.2015.7063062
DO - 10.1109/ISSCC.2015.7063062
M3 - Conference contribution
AN - SCOPUS:84940782753
T3 - Digest of Technical Papers - IEEE International Solid-State Circuits Conference
SP - 334
EP - 335
BT - 2015 IEEE International Solid-State Circuits Conference, ISSCC 2015 - Digest of Technical Papers
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2015 62nd IEEE International Solid-State Circuits Conference, ISSCC 2015 - Digest of Technical Papers
Y2 - 22 February 2015 through 26 February 2015
ER -