TY - JOUR
T1 - Low-Power MCU with Embedded ReRAM Buffers as Sensor Hub for IoT Applications
AU - Chien, Tsai Kan
AU - Chiou, Lih Yih
AU - Sheu, Shyh Shyuan
AU - Lin, Jing Cian
AU - Lee, Chang Chia
AU - Ku, Tzu Kun
AU - Tsai, Ming Jinn
AU - Wu, Chih I.
N1 - Publisher Copyright:
© 2011 IEEE.
PY - 2016/6
Y1 - 2016/6
N2 - This paper proposes embedding 256 Kb resistive random-Access memory (ReRAM) in a microcontroller unit as a data buffer for communicating with a stand-Alone flash memory. In this study, the chip was manufactured using a combination of the TSMC 0.18 \mu{\rm m} process and the Industrial Technology Research Institute ReRAM back-end-of-line process. The ReRAM was equipped with a novel sense amplifier that had three magnification times for the reference cell current for increasing the read yield by 32%. Furthermore, the ReRAM controller included built-in self-Test, built-in self-repair, a shortened Bose-Chaudhuri-Hocquenghem (99, 64, 5) error-correlating code, and asymmetric coding can yield ReRAM up to approximately 100%. When compared with the conventional dynamic random-Access memory (DRAM) buffer, the proposed architecture reduces the system execution time by 25% and the power consumption by 15% at 25 MHz. Simulations also showed that the ReRAM buffering runs at least 51% faster when compared with the use of other nonvolatile memories such as ferroelectric RAM, phase change RAM, and conduct-bridge RAM. Although ReRAM buffering is just competitive in speed and power consumption of spin-Transfer torque magnetorresistive RAM buffering, ReRAM has clear advantages in area, cost, and reliability.
AB - This paper proposes embedding 256 Kb resistive random-Access memory (ReRAM) in a microcontroller unit as a data buffer for communicating with a stand-Alone flash memory. In this study, the chip was manufactured using a combination of the TSMC 0.18 \mu{\rm m} process and the Industrial Technology Research Institute ReRAM back-end-of-line process. The ReRAM was equipped with a novel sense amplifier that had three magnification times for the reference cell current for increasing the read yield by 32%. Furthermore, the ReRAM controller included built-in self-Test, built-in self-repair, a shortened Bose-Chaudhuri-Hocquenghem (99, 64, 5) error-correlating code, and asymmetric coding can yield ReRAM up to approximately 100%. When compared with the conventional dynamic random-Access memory (DRAM) buffer, the proposed architecture reduces the system execution time by 25% and the power consumption by 15% at 25 MHz. Simulations also showed that the ReRAM buffering runs at least 51% faster when compared with the use of other nonvolatile memories such as ferroelectric RAM, phase change RAM, and conduct-bridge RAM. Although ReRAM buffering is just competitive in speed and power consumption of spin-Transfer torque magnetorresistive RAM buffering, ReRAM has clear advantages in area, cost, and reliability.
UR - http://www.scopus.com/inward/record.url?scp=84981763404&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84981763404&partnerID=8YFLogxK
U2 - 10.1109/JETCAS.2016.2547778
DO - 10.1109/JETCAS.2016.2547778
M3 - Article
AN - SCOPUS:84981763404
SN - 2156-3357
VL - 6
SP - 247
EP - 257
JO - IEEE Journal on Emerging and Selected Topics in Circuits and Systems
JF - IEEE Journal on Emerging and Selected Topics in Circuits and Systems
IS - 2
M1 - 7453196
ER -