Lead Investigators

Dustin McIntire, Timothy Chow, and William Kaiser 

Overview

A broad range of embedded networked sensor (ENS) systems for critical environmental monitoring applications now require complex, high peak power dissipating sensor devices, as well as on-demand high performance computing and high bandwidth communication.  Embedded computing demands for these new platforms include support for computationally intensive image and signal processing as well as optimization and statistical computing. To meet these new requirements while maintaining critical support for low energy operation, a new multiprocessor node hardware and software architecture, Low Power Energy Aware Processing (LEAP), has been developed.  The LEAP architecture integrates fine-grained energy dissipation monitoring and sophisticated power control scheduling for all subsystems including sensor subsystems.

Approach

A broad range of embedded networked sensor (ENS) systems for important environmental monitoring and other applications now require advanced capabilities to support high power sensor devices such as imaging devices.  Many of these applications also require support for on-demand high performance computing and communication for complex information processing.  This includes image processing, statistical computing, and optimization algorithms required for selection of proper sensor sampling.  In order to support these applications efficiently we require an ENS platform designed with integrated energy monitoring and scheduling features.

   

Figure 1: The LEAP2 platform photograph and block diagram

The LEAP2 platform is a second generation low power, energy aware processing (LEAP) ENS system that was designed specifically to allow high accuracy, low overhead energy measurement of platform resources at granularity levels previously unachievable.  Additionally, energy measurement domains may be added to or removed from the platform through an expandable energy monitoring bus integrated into the LEAP2 stacking connectors. We note that LEAP enables energy aware applications through scheduling and energy profiling of high energy efficiency components including multiple wireless network interfaces, storage elements, and sensing capabilities.  As ENS applications continue to increase in scale and complexity, energy profiling with high temporal resolution is now required to permit per process and per subsystem energy accounting. 

At the heart of the LEAP system is its Energy Management and Preprocessing (EMAP) capability.  On LEAP2 this is integrated into a dedicated ASIC implemented in a micro-power antifuse based FPGA.  The EMAP2 ASIC performs continuous, real-time energy monitoring functions as well as sophisticated power scheduling across the entire LEAP2 platform.  This allows for a new design approach that focuses on minimizing energy required for each individual sensing, computing, and communication task.  Through the EMAP2 ASIC, LEAP2 peripherals may be scheduled for use only when needed and detailed energy information is gathered during their operation.  Energy usage information for individual platform subsystems including computational resources, such as the PXA270 microprocessor, memory subsystems, such as the SDRAM, NOR flash, NAND flash and SRAM, and peripheral subsystems, such as the Ethernet, 802.11, USB, Imaging, Compact Flash, and external sensors modules are all available at millisecond accuracies.  In addition, the EMAP2 ASIC’s energy data and scheduling controls are available to the host processor through a high bandwidth memory bus interface, minimizing measurement overhead issues.  This enables the host processor to obtain energy usage information across a wide range of devices at millisecond intervals and with a minimal overhead. These features provide LEAP2 with a unique platform monitoring and control capability that allows one to significantly reduce overall platform energy usage.

Systems/Experiments

Using the real-time energy measurement capabilities of the EMAP2 ASIC, we have developed etop: a tool for application energy profiling. Based on the well-known UNIX program top, etop provides per process energy measurements with millisecond resolution. Measurements include instantaneous and average current and power draw as well total energy consumption on a per-process basis. etop also provides detailed information about the processes' energy consumption on a subsystem basis. Therefore, it is now possible to obtain a real-time and continuously updated energy profile of an entire application, consisting of one or multiple processes. Using this information, application designers can identify the major energy consuming parts of the system and adapt their algorithms accordingly.

Accomplishments

• Development of the LEAP2 platform including an Energy Management and Accounting Processor (EMAP) ASIC, which is responsible for high granularity energy monitoring and system energy scheduling
• National instruments LabVIEW interface for LEAP2
• Etop – Energy Aware Linux for the LEAP2 platform including an etop application displaying per process energy dissipation information

Future Directions

The etop demonstration on LEAP2 emphasizes how platform energy usage characteristics on high performance ENS systems is highly dynamic and varies as a result of many factors including application workload, network activity, and environmental context.  A platform with the capabilities of LEAP2 and visualization tool such as etop are critical for enabling future ENS applications with run time energy profiling data so that competing ENS applications may be explored and the most efficient algorithms chosen.

etop is part of our larger development effort to create an energy-aware operating system for embedded networked sensors.  Future plans entail integrating the operating system’s knowledge of per process energy usage into various OS subsystems such as performing energy optimized memory allocation and adaptive device power management and process scheduling based upon previous energy profiles.  By enabling the OS with detailed energy dissipation information, we can make more energy efficient choices in storage, process, networking, and sensing alternatives.

People

Faculty:

• William Kaiser, UCLA EE department
• Thanos Stathopoulos, FORTH ICS, Crete, Greece

Graduate Students:

• Dustin McIntire, UCLA EE department
• Timothy Chow, UCLA EE department

External Research Partnerships

We are currently working on an energy aware Linux kernel for the LEAP2 system in cooperation with Thanos Stathopoulos at FORTH ICS in Crete, Greece.