1. The Low Power Energy Aware Processing (LEAP) Embedded Networked Sensor System Dustin McIntire, Bernie Yip, Hing Kei Ho, Lawrence Au, Timothy Chow, and William J. Kaiser UCLA Electrical Engineering Department

2. Power Dissipation vs. Energy Efficiency High performance components often offer high efficiency …requires new architecture, enables new design approach Computing: ~ 20 - 70x Networking: ~ 10x Similar analogies in storage and sensing

3. LEAP Hardware Architecture SPM (Slauson Processor Module) EMAP (Energy Management and Accounting Preprocessor) SPI I2C V sense Shutdown x5 + 1 +- x2 Current Sensed Supply Outputs + SPM Power (2A Max) Sensor Inputs USB Host Flash USB Host Controller SDRAMPCMCIA Address/Data Sensor Voltage Ethernet JTAGGPIO SDCard/MMC MOSFET Driver 180 Pin Interboard Connector

4. LEAP Design Course Platforms LEAP Approach –Diverse sensor systems –Fine grained platform instrumentation –Highly configurable testbeds Courses Require –Usage methods that support many disciplines at the undergrad level –Rapid implementation of complex algorithms National Instruments LabVIEW interfaces for LEAP –New National Instruments program: courseware for shared development –Drag-and-drop Actions and Events –Integration of signal processing, control systems, embedded networked sensing

5. LEAP Software Tool Development:

6. The New LEAP2 Platform Dedicated ASIC performs EMAP functions Extensive current sensing and power domain controls Extendable to future modules through stacking bus

7. Address/Data Buffers LEAP2 Module Block Diagram 512MB NAND Flash GPIO/Serial Busses Ethernet CF GPS USB Host SODIMM 200 Connector uWatt FPGA mPCI WLAN CMOS Imager Current Sensors Sensor Power Switching CC2420 EMAP ASIC LEAP2 Stacking Connector Low Power Radio Module LEAP2 Stacking Connector MSP430 Current Sensors Pwr Control ADC IF Mem Bus Low Power Radio Module (LRM) Host Processing Module (HPM) EMAP2 Mini PCI Module (MPM) Sensor Interface Module (SIM) CMOS Imager Module (CIM) 8MB SRAM 64MB SDRAM 32MB NOR Flash

8. LEAP Research and Deployments New design approach based upon a component’s energy efficiency for each sensing task Energy efficiency of system components is paramount for long lived systems Energy profiling is vital for efficient use of system resources Completed and Planned deployments –Education (75+ users in class project setting) USC and UCLA –Ecosystem (acoustic localization, microclimate, multiscale sensing) –Aquatic (distributed contaminant sensing) –Seismic (rapidly deployable arrays, GEONet)