Overview

CENS science applications ranging from terrestrial ecosystem, to aquatic, contamination, and seismic systems have introduced important new requirements for ENS platforms. In particular, ENS systems for these 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 including image processing, statistical computing, and optimization algorithms required for selection of proper sensor sampling. Prior development of ENS platforms has resulted in low power systems well matched to the requirements for supporting low power sensor devices. The computing demands for such systems were matched to low data rate and low complexity sensors. However, prior ENS platforms designed to support micropower sensor devices are not adapted to system level energy minimization for a new, expanded set of ENS requirements in environmental monitoring applications, ranging from ecosystem monitoring to public health monitoring, and security applications. These applications have large sensor and instrument device power dissipation (specifically with peak power levels far in excess of the ENS node computing and communication power levels). Computing and communication demands are also advanced in order to support the on-demand processing associated with these complex sensors. However, while performance needs for new applications have increased, it is still critical to minimize system energy dissipation.

Approach

To address these diverse applications, a new design approach is required that focuses on minimizing energy required for each sensing, computing, and communication task. This leads to the new Low Power Energy Aware Processing (LEAP) multiprocessor architecture for ENS nodes. LEAP is based on hardware and software system architectural partitioning that seeks to assign the highest energy efficiency components to each ENS platform task. Through a combination of architecture advances, increased per-task energy efficiency, and required supporting algorithms, a substantial advance in processing performance and communication bandwidth is accompanied by a reduction in energy in comparison to previous platforms for typical environmental monitoring tasks.

Figure 1. LEAP ENS architecture showing the Energy Management and Accounting Preprocessor (EMAP) and its defined power domains (shown in shaded rectangles). Energy routing and data interfaces are indicated.

The LEAP architecture shown in Figure 1, has been developed to harness the use of properly scheduled, energy efficient multiprocessor components selected to achieve the lowest per task operating energy. High energy efficiency and high power components (used on demand) are assigned to a LEAP processor partition while continuously vigilant micropower components are assigned to a LEAP preprocessor partition. The Energy Management and Accounting Preprocessor (EMAP) provide fine-grained monitoring and control of energy dissipation in all ENS subsystems. Additionally it schedules operation and power delivery to sensor systems and the LEAP’s host processor. Finally, while EMAP enables the entire LEAP system to operate at micropower vigilance, it also provides event detection and triggering capability. This allows event-triggered transition to states where platform systems are available on-demand according to schedules that match application sampling requirements.

The LEAP architecture with the EMAP preprocessor further partitions ENS node subsystems into separately managed power domains supporting individual components (for example, individual sensor devices and processors). Scheduling operation within power domains enables the LEAP system to define a broad range of power modes that are then matched to environmental monitoring demands. This allows users to develop systems with application specific operating modes meeting the minimum energy required for information acquisition subject to specific sensor system and sampling requirements of each application.

The LEAP hardware architecture is combined with a software architecture providing developer access to system energy monitoring and management along with subsystem operation scheduling. Experimental verification of LEAP operation shows that this on-demand scheduling of high energy efficiency components enables algorithms that self-adapt to event behavior and may adjust operational schedules to minimize energy dissipation for a specific detection objective. These experimental results and experience with many recent users of the platform also demonstrate the convenient development path for supporting LEAP applications.

Systems/Experiments

LEAP Platform

The LEAP platform includes a processor module, shown in Figure 2, based on the PXA255 400Mhz processor and is populated with an SDRAM bank and an Intel K3 Strataflash flash bank of up to 128MB and 64MB, respectively. In addition to the processor and memory components, the processor module includes dual wireless and wireline network interfaces that are both configurable for various devices and include independent power control.

Figure 2. Processor architecture and image (left) with Preprocessor architecture and module image (right)

The EMAP preprocessor, also shown in Figure 1, utilizes a Texas Instruments MSP430F1611 microcontroller. The EMAP allows the LEAP system to be subdivided into five independently powered and isolated domains. Power is supplied to and measured for each domain via EMAP control.In addition to detailed energy monitoring, the EMAP provides a power management scheduling capability. Finally, 12bit ADC analog sensor inputs with configurable sample rate are also provided. The LEAP Processor hosts an embedded Linux operating system while the EMAP supports the uCOS real time operating system. Access to all EMAP functions and all scheduled operations is provided to applications executing on the Processor system.

LEAP Experimental System Results

The LEAP system has been characterized with a unique testbed that includes a physical event generator and distributed LEAP nodes for verification of detection, identification, and tracking problems. The physical event generator (producing mobile events consisting of sequentially illuminated and spatially distributed lamp bulbs in a test laboratory), is shown in Figure 3. LEAP nodes are equipped with micropower sensors (supported by the EMAP) and actuated sensors including camera systems. In addition to enabling fundamental investigations of energy aware algorithms in a precise, reproducible fashion, the LEAP systems and testbed have supported both undergraduate and graduate courses. Student course projects have ranged from energy aware detection and tracking of moving objects to energy aware fault detection and recovery systems that all adapt to environmental context to reduce energy. All algorithms are distributed and involve software systems operating only on the LEAP nodes.

Figure 3. The LEAP Testbed physical event generator with LEAP nodes is shown at left. A LEAP node with imager is shown in the center panel. The right hand panel shows time dependent power (solid line) and energy dissipation (dashed line) for one typical node in the network operating according to the energy-aware algorithm described in this section.

A current topic of investigation is the development of novel algorithms that seek to optimize application level energy and sensing performance through proper proactive scheduling and reactive operations. The experimental results from testbed characterization of an example algorithm are shown in Figure 3. This algorithm was developed to solve the problem of event detection and identification with the requirement that a distributed set of nodes must detect and identify an object (the moving lamp signal) and determine its color, detect its location using the imager, and finally compute velocity. This all must be accomplished while minimizing energy usage by limiting the time of operation of processor, wireless interface, and and image sensor systems. The algorithm for which results are shown in Figure 3, reactively seeks to determine the rate at which events occur and the velocity (rate at which the lamp signal moves) associated with events, then proactively schedules the operation of distributed nodes to minimize their energy usage. Figure 3 displays power and energy data from the period immediately after a test initiates at t = 0. Within 500 seconds the system has classified the environment behavior and has settled into a self-determined operation cycle where at approximately each 200 seconds this LEAP node is triggered from a sleep state for event characterization – no misdetections occur during this period. It is important to note that energy is used only episodically during servicing of the event. The large energy power excursions seen in the figure are due to imager operation. Then note that at t = 2400 seconds a change appears in the environment and a new context appears with a reduced event issue rate. Initially unaware of this change, the LEAP system detects this new context and expends energy in sensing and communication until the distributed LEAP nodes discover the new event context and again settle into a properly proactive optimized cycle of operation for t > 3000 s. This is a demonstration of capability and represents one member of a broad class of new investigations that may now be pursued.

Accomplishments

The LEAP system has been first released in 2005 and has now supported over 75 student users and researchers. The accomplishments in LEAP development include:

• Development of LEAP architecture based on high energy efficiency sensing, computation, and communication components supported by hardware and software energy monitoring and power scheduling systems.
• A LEAP testbed has been developed to test the hypothesis that high energy efficiency components, used on demand, and combined with proper scheduling will yield low energy and responsive sensing systems.
• LEAP algorithm development and experimental measurements demonstrate that low energy operation may be achieved for a wide range of environmental event detection and sensing applications.

The LEAP system is an open source hardware and software system with resources accessible here: http://cvs.cens.ucla.edu/viewcvs/viewcvs.cgi/leap/

Future Directions

The LEAP system reported here is being deployed in critical environmental monitoring systems for both static and actuated sensor networks as well as in research testbeds. A particularly important new objective is the development of energy-aware microserver networking solutions that may exploit multiple wireless interface options to achieve low latency network access along with low duty operation. The LEAP system has recently been adopted by Aevena Corporation and is now commercially available.

People

Graduate Students: Dustin McIntire; Kei Ho; Bernie Yip; Amarjeet Singh; Winston Wu
Faculty: William J. Kaiser, UCLA EE