Center for Embedded Network Sensing

A review of TEOS projects by CENS faculty and staff.

This work describes preliminary results from a 50 station broadband seismic network recently installed from the coast to the high Andes in Peru. UCLA's Center for Embedded Network Sensing (CENS) and Caltech's Tectonic Observatory are collaborating with the IRD (French L'Institut de Recherche pour le Developpement) and the Institute of Geophysics, in Lima Peru in a broadband seismic experiment that will study the transition from steep to shallow slab subduction. The currently installed line has stations located above the steep subduction zone at a spacing of about 6 km. In 2009 we plan to install a line of 50 stations north from this line along the crest of the Andes, crossing the transition from steep to shallow subduction. A further line from the end of that line back to the coast, completing a U shaped array, is in the planning phase. The network is wirelessly linked using multi-hop network software designed by computer scientists in CENS in which data is transmitted from station to station, and collected at Internet drops, from where it is transmitted over the Internet to CENS each night. The instrument installation in Peru is almost finished and we have been receiving data daily from 47 stations (out of total 49) since Jan 2009. Two remain without any network connectivity. The software system provides dynamic link quality based routing, reliable data delivery, and a disruption tolerant shell interface for managing the system from UCLA without the need to travel to Peru. The near real-time data delivery also allows immediate detection of any problems at the sites. We are building a seismic data and GPS quality control toolset that would greatly minimize the station's downtime by alerting the users of any possible problems.

The most common programming languages and platforms for sensor networks foster a low-level programming style. This design provides fine-grained control over the underlying sensor devices, which is critical given their severe resource constraints. However, this design also makes programs difficult to understand, maintain, and debug. In this work, we describe an approach to automatically recover the high-level system logic from such low-level programs, along with an instantiation of the approach for nesC programs running on top of the TinyOS operating system. We adapt the technique of symbolic execution from the program analysis community to handle the event-driven nature of TinyOS, providing a generic component for approximating the behavior of a sensor network application or system component. We then employ a form of predicate abstraction on the resulting information to automatically produce a finite state machine representation of the component. We have used our tool, called FSMGen, to automatically produce compact and fairly accurate state machines for several TinyOS applications and protocols. We illustrate how this high-level program representation can be used to aid programmer understanding, error detection, and program validation.

Data, statistical models and inferential procedures permeate CENS deployments, from the four founding scientific application areas to the more recent urban sensing campaigns. This cross-center research breaks down into three classes of research: 1) General statistical models for embedded sensing, with specific applications to data quality and continuous sampling, 2) Significant CENS-designed and supported databases and repositories, and 3) Studies into the data lifecycle for embedded sensing systems.

There exist many biological sensing applications where direct measurement is either impossible, extremely invasive, or extremely time consuming. For example, measuring the presence/absence of birds at a feeder station currently requires a human to watch a camera pointed at the feeder, identifying when birds arrive and leave. Similarly, measuring CO2 flux from a plant requires placing the plant inside a growth chamber, destructively modifying the environment. We propose using imagers as biological sensors by constructing a procedure that uses images to obtain approximate measurements of these phenomena. This procedure, composed of state-of-the-art computer vision, image processing, and statistical learning algorithms, will be evaluated in the context of a specific application and shown to be general through multiple instantiations. Through application, it has been found that many of these algorithms make unacceptable assumptions about their input. Providing accurate data to biologists and ecologists, though the appropriate modification of these algorithms, is the ultimate goal of this work.

In this poster, we consider two major themes in the design of sensor networks: data integrity, and sampling strategies. For the data integrity problem, we propose a signature-based fault detection system for identifying both intermittent faults and persistent faults. Data-dependent features using temporal, spatial, and spatio-temporal information that are useful for detecting faults are identified. These features are combined into signatures that characterize each of the different fault types. We also discuss the problem of simultaneous parameter estimation and fault detection. In this case, parameters must be estimated from a distribution that is truncated in various ways as a result of the fault detection algorithm, which can lead to biased estimates. We propose several methods to account for the bias in parameter estimates. For the sampling problem, we describe two on-going projects. The first one deals with situations where sampling as you move (using sampling paths) is more effective than contemplating sampling points. A PAR sensor riding on a NIMS 3D node is one such situation. This configuration is especially well-suited for sampling phenomena that exhibit latent geometric structure, such as light fields in forest understories. We will consider the case where the phenomena can be approximated by a piecewise-constant field and suggest a novel estimation approach when we have sample paths as observations. The second project considers the problem of finding a sampling strategy to optimize the selection of the correct regression model from a set of competing regression models. The solution is driven by minimizing the probability of error in the selection and consists of a sequential algorithm that directs the collection of measurements. We develop an adaptive sampling algorithm to sample the field with a set of static sensors and one mobile sensor. The algorithm aims to jointly minimize the probability of error in the selection and the mobility cost. The algorithm presented provides a significant improvement in the probability of error in the selection of the correct model over the random collection of measurements.

The average person is currently unaware of the real-time energy consumption for the different household appliances that he uses. At best, he can observe the monthly or bi-monthly bill indicating the total power consumption of all the appliances combined. This makes it difficult to improve the consumption efficiency, since there is no visibility in the data that he can access. We believe that real-time appliance level monitoring is necessary to allow residents to manage their energy consumption efficiently. However, monitoring end-point level power consumption is difficult to impossible with current technologies because expensive sensors, or professionally installation is necessary. In addition, device aesthetic and the inherent intrusiveness of direct in-line sensors to measure the energy usage at every end-point complicate such a system installation. Since appliances emit measurable signals when they are consuming resources, we argue that less-invasive sensors can be used for inferring real-time resource consumption. However, indirect sensors cannot be calibrated during manufacturing because of varying ambient conditions. Thus, the main challenge becomes to provide a method that autonomously calibrates the sensors. We seek to develop an easy and self-configurable monitoring system for very fine grained resource monitoring in residential spaces.

The overarching theme of the Center’s Aquatic application area continues to be the creation and application of a new genre of wireless sensing systems that will provide real-time monitoring capabilities of chemical, physical and biological parameters in freshwater and coastal marine ecosystems. High-resolution temporal and spatial measurements are essential for understanding the highly dynamic nature of aquatic ecosystems and the rapid response of microbial communities to environmental driving forces. Our unique approach to aquatic sensing and sampling, Networked Aquatic Microbial Observing Systems (NAMOS), employs coordinated measurements between stationary sensing nodes (buoys and pier-based sensors) and robotic vehicles (surface robotic boats and autonomous underwater vehicles) to provide in-situ, real-time presence for observing plankton dynamics (e.g. phytoplankton abundance, dissolved oxygen), and linking them to pertinent environmental variables (e.g. temperature, light, nutrients, etc.). Specific projects undertaken in this application area involve the development and deployment of sensor networks to examine harmful algal blooms within King Harbor, City of Redondo Beach, and the construction of mobile sensor networks in open coastal waters off southern California. The latter research involves deployments of autonomous surface and underwater vehicles, and the development of hardware and software for coordinated activities of these robotic vehicles.

Geographic Information Systems (GIS) is a powerful mapping tool that can be used to reveal spatial and temporal relationships of a criteria of interest. We have used GIS to visualize the seasonal and spatial distribution of microbial pollution obtained from the Heal the Bay beach water quality report (2007). These maps can be used to inform sampling decisions; more specifically, we can use it to identify areas of chronic pollution and can be used as a testbed for a rapid sensing system for bacteria. This rapid detection system can be used to provide higher resolution and understanding of water pollution as well as assist in understanding/characterizing environmental water quality in specific areas. We propose the subsequent use of an covalently-linked immumomagnetic separation/ATP quantification assay that is rapid, robust, and field-portable as an instrument to conduct monitoring of E. coli and Enterococcus in marine and freshwater systems.

CENS research related to developing and implementing structural health monitoring (SHM) systems is advancing on two distinct but related fronts: ShakeNet, a portable wireless sensor network for rapid, post-event deployments and SHMnet, a novel SHM system for permanent monitoring of tall buildings and special structures in Los Angeles. The primary objective of the SHMnet research is the development of a robust SHM system along with the associated hardware and software, using tall and special structures (e.g., bridges, port structures, dams) in Los Angeles as a testbed. More specifically, the development of a wireless Data Acquisition (DAQ) toolbox suitable for rapid urban deployments, a suite of state-of-the-art sensors for monitoring key structural responses including innovative methods for directly measuring interstory displacements, and probabilistic post-event assessment algorithms based on experimental motion-damage relationships. Progress on these fronts is highlighted. One rather unique aspect of this research stems from partnerships with strong-motion instrumentation programs (SMIPs) such as CSMIP, ANSS, and the LA-DBS. The proposed SHMnet leverages both building access and instrumentation requirements already facilitated by one or more SMIPs. However, a critical look at structural instrumentation guidelines of various SMIP agencies exposed a lack of uniformity of experience-based specifications. To this end, we sought to establish a quantitative basis for key structural instrumentation specifications, namely sample rate, resolution, and time synchronization. This was accomplished by analyzing signal errors associated with data acquisition processes and engineering sensitivity analyses of several intensity measures and engineering demand parameters. Results from these studies will be useful in updating current structural instrumentation specifications of major SMIPs as well as provide specifications for SHMnet tools. ShakeNet is a portable wireless sensor network for instrumenting large civil structures such as buildings and bridges. The focus of ShakeNet design is to take advantage of wireless technology for deployments in structural environments where power or communications infrastructure is nonexistent or unavailable. It is designed to collect structural vibration measurements for up to a week from each node within the network by deployment in large structures within hours after an earthquake. It will consist of 25 sensor nodes and 5 to 10 master-tier nodes (Stargates or other embedded computers) that provide increased communications capacity. The ShakeNet software subsystem is built upon Tenet; programmable wireless sensing software designed for multi-tier sensor networks. ShakeNet will be deployed and tested on several structures that represent a range of structure types, environments, ages, and degrees of retrofit. They include the Seven Oaks Dam in Redlands, CA, the Santa Ana River Bridge in Riverside, CA, 1100 Wilshire Blvd. in downtown Los Angeles, CA, and the Long Beach Veterans Administration Hospital in Long Beach, CA.