Ultra-low power Technologies for “always-on” sensor node applications
IEEE SSCS/CAS Seminar
Date and Time
Location
Hosts
Registration
- Date: 16 Nov 2018
- Time: 01:00 PM to 02:00 PM
- All times are (UTC-06:00) Central Time (US & Canada)
- Add Event to Calendar
- 2501 Speedway
- Austin, Texas
- United States 78712
- Building: EER North Tower
- Room Number: EER 3.646
- Click here for Map
- Starts 01 November 2018 07:14 AM
- Ends 16 November 2018 12:00 PM
- All times are (UTC-06:00) Central Time (US & Canada)
- No Admission Charge
Speakers
Dr. Mahesh Mehendale
Ultra-low power Technologies for “always-on” sensor node applications
Abstract: Ultra-low power continues to be one of the key requirements for battery operated sensor node applications such as Industrial IoT. Duty-cycling is one of the most commonly used technique to drive longer battery life. However, the requirement for the sensor nodes to wake-up asynchronously in response to an event of interest (“always-on” operation) makes it challenging to reduce power in the standby mode. Similarly, the need for the nodes to be increasingly intelligent, makes it challenging to reduce power in active mode. In this talk, we will discuss low power technologies addressing both the standby mode as well as active mode power reduction in “always-on” intelligent sensor nodes. We will highlight how it is important to change the paradigm from ‘focusing on reducing power consumption alone’ to ‘driving energy efficiency across power conversion, delivery, storage and consumption’. In that context, we will discuss the need to drive ultra-low Iq power management solutions which can support a wide range (>6 orders of magnitude) load currents. Instead of sending the raw data from sensors to the cloud for analysis, the “intelligent” nodes process the data locally and drive local decisions, so as reduce latency and data communication bandwidth. As the machine learning technologies mature, there is an opportunity to move them from the cloud and the gateways to the end nodes. In the talk, we will discuss some early ideas on realizing neural networks on sensor nodes under aggressive cost and energy constraints. We believe that mixed-signal approaches and techniques such as in-memory compute are critical to achieve an order of magnitude energy reduction needed by these applications.
Biography:
Bio: Mahesh Mehendale is a TI Fellow and leads the Nano-power Foundational Technology at Kilby Labs. His current areas of focus include ultra-low power circuits and micro-architectures for “always-on” sensor nodes. Before this, he worked on architectures for low-power micro-controllers and high-definition (HD) video compression. Since joining TI in 1986, he has led the development of multiple industry-leading digital and system-on-chip (SoC) designs, including C27x/C28x DSPs and DM642 digital media processor. Mahesh has published more than 50 papers at international conferences/journals and presented many invited talks/tutorials. He has co-authored a book on "VLSI synthesis of DSP kernels" and two book chapters. Mahesh holds eight U.S. patents with 8 more applications pending, and was elected senior member of IEEE in 2000. He received the "Distinguished Alumnus" award from Indian Institute of Technology (IIT) Bombay in 2012, the Zinnov Award for Thought Leadership in 2014 and was elected Fellow of the Indian National Academy of Engineers in 2016.