Overview

CENS extensive field experimentation has uncovered new requirements for characterizing previously unknown phenomena. The complexity and dynamic nature of the observed phenomena present measurement challenges that are well beyond the capabilities of prior methods.  Rapid spatiotemporal variation, constraints demanding rapid deployment, and the requirements for mapping phenomena over large areas lead to new ENS system research objectives and metrics that include: fidelity of observation, ability to share resources, rapidity of deployment, timeliness of observation result, and deployment parsimony. Multiscale actuation methods, described in this report, offer algorithms and systems that enable system optimization with respect to these observation metrics.  These require the introduction of a sensing and actuation hierarchy that augments the static fixed sensor network.  Our results of this last year demonstrate that the early methods are quite general in capability with diverse applications being supported by a common set of multiscale actuation architectures. Progress has been made in theory, software systems, and hardware platforms.

In these systems, sensors in successively higher layers contribute a broader field of view at the expense of lower resolution.  These provide a dramatic advance by guiding both the initial deployment of ENS networks and augmenting the actuation of new ENS nodes providing higher resolution sensor and/or different sensing modes. At the same time there are new challenges in system modeling and exploiting the new system capabilities to efficiently sample the environment while verifying proper operation. Many accomplishments in fundamental theory and algorithm development were made in this year, including multi-scale image classification, adaptive sampling and reconstruction, fundamental relationships between fidelity and network lifetime, coordinated actuation, fault detection and calibration, and 3-D localization.  In addition, new multiscale actuated sensing architectures were also developed. The deployment of these first systems resulted in important observations regarding both contaminant distribution and phenomena in terrestrial ecosystems.

Research Objectives

Multiscale actuated sensing research in this year has been directed to the objectives for developing: 1) ENS software and platform systems that exploit actuation and multi-tier sensor assets to properly guide sampling for a wide range of applications, 2) ENS software and platforms that enable high spatiotemporal resolution monitoring, 3) Localization and actuated sensing methods that exceed the performance of prior fixed sensor systems, and 4) fault detection and calibration methods that exploit actuation for diverse sensor systems and applications.

Progress has appeared in each of these areas and forecasts further advances in the future for both research in multiscale actuated sensing and applications. For example, to study microbial abundances, stationary NAMOS buoys have been developed that measure temperature, chlorophyll concentration, wind and water speeds while the boat, equipped with GPS, water sampler, thermometer, and fluorometer, and produce detailed maps of surface chlorophyll abundance in certain regions.

Multiscale sensing systems have also been directed to the characterization of forest understory light.  Our application research in Terrestrial Ecosystems has indicated the urgency of this application as well as the inadequacy of past methods. This report describes both theory and experimental results demonstrating the advanced capabilities of multiscale methods for light field sampling and reconstruction. Localization and sensing methods have also provided systems that actively detect and reduce the impact of sensing uncertainty due to obstacles. Finally, progress has also appeared in fundamental theory, algorithms, and experimental systems in fault detection and calibration. Here actuated sensing methods have been applied to applications in Contaminant Monitoring and Management using in situ and autonomous active sensor calibration methods.   The combined progress has now created a multiscale actuation foundation for ENS systems that meet the observation metrics described above, with actuation systems ranging from constrained actuation or free-roving devices, and applications that vary widely in disciplines with variable phenomena characteristics and sensing modalities.

This multiscale actuated sensing report begins with a description of the new architectures for multiscale actuated sensing in Section A including an analysis comparing the performance of Multiscale Actuated Sensing methods with prior techniques for challenging applications in the sampling of dynamic fields.  Section B follows with a description of the problem of phenomena characterization. While applicable to a broad class of field variable sampling, this method is applied to actual data from forest ecosystem studies of primary interest to TEOS research.    Section C, further describes an important investigation of the coupled problem of model and data uncertainty that will influence the design of new model-driven sampling and field reconstruction systems.  Adaptive sampling methods, remaining at the core of CENS sampling technologies are described in Section D.  The problem of sampling is not only influenced by the constraints of sensing fidelity, but, also the requirements for optimization of resource costs, in particular that of energy.  Section E describes an investigation of the relationship between sensor system energy use (sensing system lifetime) and distortion in measurement and phenomena reconstruction resulting from reduced sampling density. The NIMS technologies that support both verification of the above algorithms and methods as well as providing science return for applications are discussed in Section F.

This report continues with Section G that discusses a theoretical and experimental investigation where multiple actuated sensors are coordinated using an approach that achieves both adequate spatial coverage, and through proactive methods, adequate temporal response.  The related visualization methods applied as well to subsurface soil studies are included in Section H.  These combined methods finally also are applied to the verification of microclimate sensors deployed in the field. Section I describes new methods for in field verification of sensing systems.  Solutions for the companion problems of in situ calibration are described in Section J.  These fundamental investigations are followed by research in actuated sensing solutions for the fundamental sensor node localization problem of Section K. This report section concludes with a description of the NAMOS system that integrates multiple tiers of fixed and actuated sensor nodes with applications to three-dimensional characterization of marine and aquatic systems.