Skip to content. Skip to main navigation.

Emerging Memory Subsystems for Internet-of-Things: Energy-efficient and Secure Information Processing and Storage

Friday, April 28, 2017, 11:00 AM - 12:30 PM
Nedderman Hall Room 203

Add to Calendar

Seung H. Kang, Ph.D.
Director of Engineering, Corporate R&D, Qualcomm Technologies, Inc.

Abstract:
Emerging Internet-of-Things (IOT) demands an energy-efficient network of smart nodes (semiconductor devices coupled with sensors) connected to gateways and the cloud that serves big data. Such nodes, distributed widely in various form factors, need to be always-on, always-aware, and always-connected, despite the fact that their active duty cycles are low (<~1%), leaving them predominantly on a standby mode. In addition, these systems are expected to be secure and available at low costs. Accordingly, these create both challenges and opportunities for nonvolatile integrated circuits, in particular, embedded nonvolatile memories (eNVM). Owing to intrinsic limitations of conventional memory technologies, present IOT systems rely both on a nonvolatile storage and on a volatile working memory. In a typical IOT cycle, the execution code is read from the NVM directly or indirectly through the working memory. In addition, the transitory data pertaining to the active cycle need to be written to and read back from the working memory. When the duty cycle is low, as in the case of IOT devices, the energy consumption can be dominated by moving the code from NVM, moving the data to NVM, and retaining at least a portion of the working memory (SRAM) in standby mode between power cycles. A unified memory subsystem that combines these two types of memories can make such transactions unnecessary and improve the overall energy efficiency dramatically. This can be realized with an emerging memory such as STT-MRAM, a new class of NVM, which is capable of fast and practically unlimited read and write operations. Furthermore, such a unified memory subsystem can provide persistency, atomicity and anti-tearing which are much desired attributes for secure information transactions, mitigating vulnerabilities of certain IOT devices.

Bio:
Seung KangSeung Kang is director of engineering, Corporate R&D, at Qualcomm Technologies, Inc., and leads an emerging memory technology group for mobile systems, IOT, and healthcare/bioelectronics. He earned B.S. and M.S. degrees from Seoul National University and his Ph.D. in materials science and engineering from the University of California at Berkeley. From 1998-2005, he worked at Lucent Technologies Bell Laboratories and led advanced device reliability projects as Distinguished Member of Technical Staff. He joined Qualcomm in 2006 and has pioneered embedded STT-MRAM and spintronic devices for mobile systems. He has served at numerous technical committees. He has delivered more than 50 keynote and invited speeches at international conferences and published 100 papers. Kang previously worked at Lawrence Berkeley National Laboratory on SQUID sensors and VLSI interconnects. He holds more than 500 patents granted globally and currently serves as an IEEE Electron Device Society Distinguished Lecturer. 

Return to Upcoming Events

 ∧