Title :
A 15µW 16 channel E×G processor with data transition memory-quad level vector for wearable healthcare platform
Author :
Roh, Taehwan ; Lee, Seulki ; Yoo, Hoi-Jun
Author_Institution :
Dept. of Electr. Eng., KAIST, Daejeon, South Korea
Abstract :
A low power 16 channel E×G processor (E×P) is proposed for wearable healthcare system. It adopts two low power accelerators named data transition memory quad level vector (DTM-QLV) and coefficient interleaving FIR filter (CI-FIR). The proposed DTM-QLV reduces transmission power consumption by compressing the data size of 4 different kinds of ExG signals. It can decrease the memory capacity by 13% compared with the previous QLV, and results in power reduction by 1/80 of the conventional RISC processor. Moreover, the proposed CI-FIR can achieve 125 times power efficient operation by multiply-less architecture compared to the RISC processor. In addition, adaptive buffer-controlled gating (ABG) helps to operate only when signal should be processed so that the average power consumption can be significantly decreased to less than 15μW for 16 channel operation.
Keywords :
FIR filters; biomedical electronics; data communication; health care; low-power electronics; medical signal processing; microprocessor chips; power aware computing; CI-FIR; DTM-QLV; ExG signal; adaptive buffer-controlled gating; coefficient interleaving FIR filter; data transition memory-quad level vector; low power 16 channel ExG processor; low power accelerators; memory capacity; multiply-less architecture; power 15 muW; transmission power consumption; wearable healthcare platform; Biomedical monitoring; Electrocardiography; Electroencephalography; Finite impulse response filter; Medical services; Power demand; Reduced instruction set computing;
Conference_Titel :
Biomedical Circuits and Systems Conference (BioCAS), 2011 IEEE
Conference_Location :
San Diego, CA
Print_ISBN :
978-1-4577-1469-6
DOI :
10.1109/BioCAS.2011.6107793