Battery Research at Microsoft

Mobile & Internet of Things (IoT) devices, along with other battery operated devices, are energy constrained. While hardware capabilities have increased tremendously over the last ten years, battery energy density has only doubled. In this project we are exploring several techniques to extend the battery life of mobile and IoT devices.

Our work spans innovation across all layers of the battery stack – from new battery chemistries to power management hardware, from device drivers and operating system abstractions to the energy-efficient apps that run on the device. We work closely with product teams across Microsoft including the Windows, Surface and HoloLens teams which provides a very real context for our longer-term research agenda which aims to change the way we think about batteries.

Highlights

  • November 2017: Read about SDB implementation in Windows here
  • SDB covered by the press (MIT Technology Review, The Register, Yahoo Finance)
  • Microsoft Research blog on Software Defined Batteries (Article)
  • WearDrive wins Best Paper Award at USENIX ATC 2015 (Press Article)
  • SDB paper was accepted for inclusion at SOSP 2015
  • Paper on reducing energy consumed by Wi-Fi accepted in UbiComp 2015
  • WearDrive paper was accepted for publication at USENIX ATC 2015
  • Wireless Display energy consumption paper accepted to MobiSys 2015
  • ComputerWorld highlights our energy work at MSR (Press Article)
  • XBOX One wireless controllers are most energy efficient game controllers in the market!
  • Visual Studio tool for estimating energy consumption by apps highlighted by Steve Ballmer in 2013 Build Keynote (around 45th minute)
  • Research

    While we continue to make Microsoft products energy efficient by designing and incorporating new energy saving techniques, we are also inventing fundamentally new technologies with a view to achieving the next breakthrough in mobile device and wireless sensor battery life.

    Offload: We have developed techniques which allow a system to dynamically move computation, storage and/or networking operations to a more energy efficient component of a system. This could be a lower power processor on the phone, or a less battery constrained device, such as a router, nearby PC, or the cloud. We have developed techniques to offload storage (WearDrive), and computation to a low-power processor (Somniloquy, LittleRock), or to a nearby PC or cloud (MAUI).

    Software Defined Batteries: Different battery chemistries perform better on different axes such as: energy density, cost, peak power, recharge time, longevity and efficiency. Mobile system designers are constrained by existing technology, and are forced to select a single chemistry that best meets their diverse needs, compromising other desirable features. We have developed a new hardware-software system called the Software Defined Battery (SDB) that allows system designers to integrate batteries of different chemistries. SDB exposes APIs to the operating system to control the amount of charge flowing in and out of each battery enabling it to dynamically trade one battery property for another.

    Technology Transfers

    Low Energy Wi-Fi (2012) – Windows 8, and later versionsIncreases battery lifetime in Windows 8 Tablets and Surface computers.

    Compared to laptops the new class of mobile devices, such as tablets and Surface computers, need to stay connected even when the screen is turned off. Keeping the Wi-Fi always on consumes significant energy. We designed a set of techniques that allows the Wi-Fi device to not lose its connection even when the screen is turned off and the processor (and SoC) is in a low power state. We accomplished this by reducing the Wi-Fi power consumption to a few mW in standby state. Our techniques shipped in Windows 8, and later versions of Windows.

    Visual Studio Energy Modeler & Profiler (2012) – Windows Phone 8 Dev ToolsEnables mobile app developers to estimate the energy consumption of their apps

    Poorly written apps are one of the primary reasons for high energy drain on mobile devices. One reason for energy-inefficient apps is that app developers do not have sufficient tools to determine the energy impact of their apps. As part of a Wattson research project we designed a Visual Studio plug-in that provides visibility to the application developer of their application’s energy consumption. Our paper Empowering Developers to Estimate App Energy Consumption, published in ACM MobiCom 2012 describes the details of the system. This work formed the basis for the Energy Profiler that is part of the Visual Studio SDK for Windows Phone 8.

    Visual Studio 2013 Energy Profiler (2013) – Diagnostics & Performance Toolbox for Windows 8.1Enables app developers to estimate the energy consumption of their apps

    Described here, and also showcased in Steve Ballmer’s Build Keynote (around 45th minute). Our research paper Empowering Developers to Estimate App Energy Consumption, published in ACM MobiCom 2012 describes the details of the system.

    XBOX One Wireless Controller Protocol (2013)The wireless protocol between the XBOX One controller and console

    Designed the low-energy, high-throughput, low-latency wireless protocol for the XBOX One. Research and industry has always focused on getting higher throughput. However, little prior work, if any, looks at achieving high throughput while achieving low (guaranteed) latency, and consuming less energy. For the XBOX One we designed a protocol that achieves both. I enjoyed being deeply involved with the design, development, and testing of the controllers.

  • Press

  • Talks

  • People

    Left to Right (Anirudh Badam, Ranveer Chandra, Julia Meinershegan, Bodhi Priyantha)

    Not in picture:

    Interns

    • Xia Zhou (2011, UCSB, now Dartmouth): Worked on low power Wi-Fi (Shipped features in Windows 8, 8.1, and 10)
    • Apurv Bhartia (2011, UT Austin, Now Meraki): Low power Wi-Fi Display
    • Parya Moinzadeh (2012, UIUC, now Google): Software Defined Batteries & e-Loupe
    • Pan Hu: (2014, UMass): SDB Hardware
    • Tony Ferresse (2014, UC Berkeley, Now Tesla): SDB Batteries & Simulations
    • Valia Skiani (2014, Columbia): SDB Algorithms
    • Jing Li (2013, UCSD): Energy consumed by storage
    • Jian Huang (2014, Georgia Tech): WearDrive
    • Christopher Dekmezian (2015, UC Berkeley): New dynamic batteries

人员

Ranveer Chandra的肖像

Ranveer Chandra

Managing Director, Research for Industry

Bodhi Priyantha的肖像

Bodhi Priyantha

Researcher

Thomas Moscibroda的肖像

Thomas Moscibroda

Distinguished Engineer Azure Core Platform Capacity & Efficiency Microsoft Azure

Steve Hodges的肖像

Steve Hodges

Senior Principal Researcher

Riyaz Pishori的肖像

Riyaz Pishori

Principal Research PM