Ultra-Low Power Computing Workshop 2014

Advances in semiconductor, sensing, communication, energy harvesting, and energy storage continue to push the boundaries of low power devices giving rise to computing systems with extremely low power consumption. These great advances present us with important, new research challenges related to efficient system design, programming, and user interaction. This workshop lets a diverse group of researchers present their latest work in this area and discuss the future of ultra-low-power computing.

Agenda

 

Time

Event/Topic
 

8:30 – 9:00

Breakfast

 

9:00 – 10:00

Keynote by Doug Burger

  10:00-10:15     Coffee Break 

Ambiently

Powered

Devices 

10:15 – 10:30

Ranveer Chandra

10:30 - 11:00  Josh Smith 
11:00 - 11:30  Prabal Dutta 
11:30 - 12:00  Deepak Ganesan 
 

12:00 – 1:00

Lunch

Sensing

Systems 

1:00 - 1:15  Bodhi Priyantha
1:15 - 1:45     Pei Zhang 
1:45 - 2:15  David Brooks 
2:15 - 2:45  Kevin Fu 
 

2:45 - 3:00

Coffee Break

Devices &

Cloud 

3:00 - 3:15

Jie Liu

3:15 - 3:45  Ras Bodik 
3:45 - 4:15  Ben Lee 
 

4:15 - 4:25

Board Bus to Social Event

  4:25  Depart for Black Bottle Postern 
  4:45 - 7:00  Reception at Black Bottle Postern 

 

Speakers

  • Keynote by Doug Burger (Microsoft Research)
  • Ras Bodik (University of California, Berkeley)
  • David Brooks (Harvard University)
  • Ranveer Chandra (Microsoft Research)
  • Deepak Ganesan (University of Massachusetts, Amherst)
  • Benjamin Lee (Duke University)
  • Jie Liu (Microsoft Research)
  • Bodhi Priyantha (Microsoft Research)
  • Josh Smith (University of Washington)
  • Pei Zhang (Carnegie Mellon University)
  • Kevin Fu (University of Michigan)
  • Prabal Dutta (University of Michigan)

 

Detailed Agenda

 

Keynote by Doug Burger - Microsoft Research

 

Ras Bodik - University of California, Berkeley

Constructing programming models and compilers for minimalistic low-power architectures

Abstract: In the name of energy efficiency, low-power hardware will shed many programmability features, increasing the abstraction gap. The burden of bridging the gap will fall on compiler writers who are not well equipped to write compilers from novel programming models to idiosyncratic architectures because traditional rewriting transformations won't be applicable and the non-traditional ones may take years to develop.

I will describe how a synthesis-aided compiler can sidestep code transformations, enabling compiler prototyping while producing highly efficient code. As a case study, I will describe Chlorophyll, a synthesis-aided programming model and compiler for the GreenArrays GA144, a minimalist low-power spatial architecture that requires partitioning the program into fragments of no more than 256 instructions and 64 words of data. The talk is based on our paper in PLDI 2014

Bio: Ras Bodik is a Professor of Computer Science at University of California, Berkeley. He works on foundations and applications of program synthesis. His foundation projects include solver-aided languages, synthesis-aided human-computer interactions, and modular synthesis. His group is also rethinking the compiler architecture, replacing rewrites with simple synthesis algorithms. Their synthesis-aided compilers produced parallel document layout engines and targeted ultra-low-power architectures. He also works on computational doing, which brings synthesis tools to the hands of scientists, with applications to systems biology and working with big data.

 

David Brooks - Harvard University

The Harvard Robobees Project: A Convergence of Body, Brain, and Colony


Abstract: The Harvard Robobees Project (http://robobees.seas.harvard.edu) is a large-scale multidisciplinary project that seeks to develop a swarm of autonomous, life-sized robotic insects. The driving application for small scale mechanical flying insects is to autonomously pollinate a field of crops, while other applications range from search-and-rescue to mobile sensing. Construction of these devices provide many challenges that span computer science, mechanical engineering, electrical engineering, and biological sciences. This talk will provide an overview of the three major aspects of the project: the Body, Brain, and Colony. The talk will explore accelerator-based computing architectures necessary to provide the energy/performance requirements for the Robobee Brain in a highly-integrated form factor.

Bio: David Brooks is a Gordon McKay Professor of Computer Science in the School of Engineering and Applied Sciences at Harvard University. He joined Harvard in 2002 after spending one year as a research staff member at IBM T.J. Watson Research Center. Prof. Brooks received his BS in Electrical Engineering at the University of Southern California and MA and PhD degrees in Electrical Engineering at Princeton University. His research interests include technology-aware computer design, with an emphasis on power-efficient computer architectures and system software for high-performance and embedded systems.

 

Prabal Dutta - University of Michigan

From Pervasive to Perpetual Computing: Realizing the Next Computing Class

Abstract: Every decade, computers scale in size, power, and deployment density. This talk will explore some of the opportunities and challenges in scaling to the next generation and beyond. Drawing from our experiences in deploying hundreds of energy harvesting sensors, from Rubik’s Cube-size devices that draw milliamps to cubic-mm sized that draw nanoamps, this talk will discuss the emerging systems architecture, and articulate some of the hardware and software challenges, of perpetual computing—of intermittently-powered systems that could conceivably last forever.

Bio: Prabal Dutta is an Assistant Professor of Electrical Engineering and Computer Science at the University of Michigan, Ann Arbor. He researches the circuits, systems, and software necessary to realize pervasive sensing, computing, and communications at scale and in the service of society. His work was recognized with an Intel Early Career Award and NSF CAREER award. He earned a Ph.D. in Computer Science from the University of California, Berkeley in 2009, where he designed and deployed some of the largest academic sensor networks, and where his work was supported by NSF and Microsoft Graduate Fellowships.

Kevin Fu - University of Michigan

Intentional Electromagnetic Interference

Abstract: Computers trust their inputs too much. This is especially true for sensors that convert analog signals into digital representations because such components are out of sight and out of mind. How does a sensor system know that what it sensed is real? In this talk, I will explain recent security research on how intentional electromagnetic interference can affect control of embedded control systems ranging from phone calls to pacemakers. Based on joint work published at IEEE Security & Privacy with Denis Foo Kune, John Backes, Shane S. Clark, Daniel Kramer, Matthew Reynolds, Yongdae Kim, and Wenyuan Xu.

Bio: Kevin Fu is Associate Professor of EECS at the University of Michigan where he directs the Archimedes Center for Medical Device Security and the SPQR group. His research investigates how to achieve trustworthy computing on embedded devices with application to health care, commerce, and communication. Kevin received his PhD in EECS from MIT. He received a Sloan Research Fellowship, NSF CAREER award, Fed100 Award, and best paper awards from various academic silos of computing. Kevin was named MIT Technology Review TR35 Innovator of the Year for work on medical device security. Kevin has testified in Congress on health matters and has written commissioned work for the Institute of Medicine of the National Academies. He served as a visiting scientist at the U.S. Food & Drug Administration, the Beth Israel Deaconess Medical Center of Harvard Medical School, Microsoft Research, and MIT CSAIL. He is a member of the ACM Committee on Computers and Public Policy and the NIST Information Security and Privacy Advisory Board. Kevin also holds a certificate of achievement in artisanal bread making from the French Culinary Institute.

 

Deepak Ganesan - University of Massachusetts, Amherst

Ultra-low power sensing via backscatter communication and sparse sampling

Abstract: While high-rate sensors such as cameras are commonplace on mobile and wearable devices, they are too power-hungry for continuous operation on batteries and require either tethering or frequent recharges. This is not surprising — such sensors are often power-hungry, have high data handling needs, and require substantial bandwidth, all of which makes them difficult to optimize for power consumption.

In this talk, I will make the case for re-thinking the design of high-rate wireless sensors by leveraging backscatter communication and sparse sampling. From a communication perspective, I will discuss how we can address range and throughput limitations of backscatter to enable high-rate wireless backhaul at micro-watts. From a sensing perspective, I will discuss how application-driven sparse sampling can dramatically reduce sensing costs, and demonstrate the use of this approach to design ultra-low power computational eyeglasses that continually tracks eye and visual context.

Bio: Deepak Ganesan is Associate Professor in the Department of Computer Science at UMASS Amherst. He received his Ph.D. in Computer Science from UCLA in 2004 and his bachelors in Computer Science from IIT, Madras in 1998. He has received the NSF CAREER Award, the IBM Faculty Award, and a UMass Lilly Teaching Fellowship. Most recently, his work recently received a Best Paper Award at ACM CHI 2013, and an Honorable Mention for Best Paper Award at ACM Ubicomp 2013. He was a Program co-chair for ACM SenSys 2010 and IEEE SECON 2013.

 

Benjamin C. Lee - Duke Univeristy

Economic Mechanisms for Managing Risk in Heterogeneous Datacenters

Abstract: As cloud computing proliferates, demand for datacenter computing capacity increases. Moreover, we must increase capacity within today's megawatt-scale power budgets. Toward this goal, we present the case for building datacenters using processors and memories that were originally intended for mobile and embedded platforms. For web search, mobile processors are 5x more efficient than server processors. We quantify and mitigate the impact on query latency, relevance, and quality-of-service. Mobile memories are 5.6x more efficient than server memories. We identify datacenter applications that can benefit from mobile memories.

Mixing server and mobile hardware in a datacenter increases management complexity and we describe how datacenters might navigate this complexity with economic mechanisms. For settings where throughput is desired, we present a market in which users bid for heterogeneous hardware. For settings where fairness is desired, we present a game-theoretic mechanism that guarantees equitable hardware allocations.

Bio: Benjamin Lee is an assistant professor of Electrical and Computer Engineering at Duke University. His research focuses on scalable technologies, power-efficient architectures, and high-performance applications. He is also interested in the economics and public policy of computation. He has held visiting research positions at Microsoft Research, Intel Labs, and Lawrence Livermore National Lab.Dr. Lee received his B.S. at the University of California at Berkeley, S.M. and Ph.D. at Harvard University, and post-doctorate at Stanford University. He received the NSF CAREER Award in 2012. And his research has been honored as a Top Pick by IEEE Micro Magazine (2010), twice as a Research Highlight by Communications of the ACM (2010, 2011), and by an NSF Computing Innovation Fellowship (2009-10).

 

Joshua Smith - University of Washington

RF-Powered Sensing, Computing, and Communication

Abstract: The ability to deliver power using Radio Frequency (RF) signals instead of wired connections is changing the design of systems such as implanted medical devices, smart structures, and consumer electronics. From Eniac to today, the energy efficiency of computing has improved by a factor of one trillion; this energy scaling trend is one of the key enablers for wireless power today.I will describe several systems my lab has developed in the course of exploring the space of wireless power techniques. WISP (Wireless Identification and Sensing Platform) is a platform for sensing and computing that is powered and read by standards-compliant UHF RFID readers. It has been used by researchers around the world for both "perpetual sensing" and RFID security research. Our ABC (Ambient Backscatter Communication, joint work with Shyam Gollakota) nodes are powered by ambient RF signals (such as TV or cell phone signals); ABC nodes communicate by reflecting pre-existing RF signals, rather than generating their own signals.WISP and ABC operate in the far field and have power budgets of tens of microwatts. WREL (Wireless Resonant Energy Link) and FREED (Free-range Resonant Electrical Energy Delivery) operate in the near field and provide tens of watts. The FREED system powers Left Ventricular Assist Devices (LVADs), implanted heart pumps that today require a transcutaneous power cable. I will also outline our recent projects on fully implanted ElectroCorticography and Brain-Computer-Spinal Interface. I will conclude by reflecting on changes to the design space enabled by wireless power, and discussing future directions for RF-powered sensor and actuator systems.

Bio: Joshua R. Smith is an Associate Professor in the departments of Computer Science and Engineering and Electrical Engineering at the University of Washington, Seattle, where he leads the Sensor Systems research group. He is an Allen Distinguished Investigator, he is the thrust leader for Communications and Interface in the NSF Engineering Research Center (ERC) for Sensorimotor Neural Engineering, and he is the theme leader for Low Power Sensing and Communication in the Intel Science and Technology Center for Pervasive Computing.In recent years his research as focused on wirelessly powering and communicating with sensor systems in applications such implanted biomedical electronics, ubiquitous computing, and robotics. Previously, he co-invented an electric field sensing system for suppressing unsafe airbag firing that is included in every Honda car. He is the editor of a book entitled “Wirelessly powered sensor systems and computational RFID” (Springer, 2013) that includes his work in this area as well as related work by other researchers. He received B.A. degrees in computer science and philosophy from Williams College, the M.A. degree in physics from Cambridge University, and the Ph.D. and S.M. degrees from the MIT Media Lab’s Physics and Media group.

 

Pei Zhang - Carnegie Mellon University

Air-Borne and Body-Borne Real-time Sensing and Processing

Abstract:  As sensors become more ubiquitous in the environment, they are increasingly relied upon to in many applications. These sensors provide a rich kaleidoscope of physical data that was not previously available. This data is forcing us to rethink traditional sensing systems and how they function. This talk will explore this issue of obtaining needed sensing information indirectly from a set of noisy sensors in physical sensing systems. The talk first introduces the SensorFly system aimed at emergency situations. SensorFly is a novel, low-cost, miniature controlled-mobile aerial sensor network weighing only 30 grams each. This limits the capability of each device but permits indirect sensing through in-network understanding of motion data. The talk then move to incorporate other sensing modalities into personal and wearable systems, where sensing data is extracted indirectly from a multitude of sensors.

Bio: Pei Zhang is an associate research professor in the INI, Silicon Valley and ECE departments at Carnegie Mellon University. He received his Bachelor's degree with honors from California Institute of Technology in 2002, and his Ph.D. degree in Electrical Engineering from Princeton University in 2008. While at Princeton University, he developed the ZebraNet system, which is used to track zebras in Kenya. It was the first deployed, wireless, ad- hoc, mobile sensor network. Focusing on collaborative and pervasive systems utilizing groups of sensors, his recent work includes SensorFly (focus on groups of autonomous miniature-helicopter based sensor nodes) and MARS (Muscle Activity Recognition). Beyond research publications, his work has been featured on popular media including CNN, Science Channel, Discovery Channel, CBS News, CNET, Popular Science, BBC Focus, etc. He is also a cofounder of Vibradotech. In addition, he has won several awards including the NSF CAREER, Edith and Martin B. Stein Solar Energy Innovation Award, and a member of Department of Defense Computer Science Studies Panel.

  

 

Logistics

Venue

Microsoft Conference Center Building 33 16070 NE 36th Way, Redmond, WA 98052.

Date

July 16, 2014

Register for the event

Visit Information

Microsoft Research Redmond visit information

People
Brandon Lucia
Brandon Lucia

Jie Liu
Jie Liu

Arjmand Samuel
Arjmand Samuel

Desney Tan
Desney Tan