This tutorial presents a detailed study of sensor faults that occur in deployed sensor networks and a systematic approach to model these faults. We begin by reviewing the fault detection literature for sensor networks. We draw from current literature, our own experience, and data collected from scientific deployments to develop a set of commonly used features useful in detecting and diagnosing sensor faults. We use this feature set to systematically define commonly observed faults, and provide examples of each of these faults from sensor data collected at recent deployments.

The identification of sensors returning unreliable data is an important task when working with sensor networks. The detection of these unreliable sensors while in the field can cue human involvement in repairing problem sensors. This ensures that meaningful data is collected throughout the entire length of a sensor deployment. We present a detection based method of identifying faulty and non-faulty sensors from a given set of sensors that are expected to behave similarly. We use a Bayesian detection approach to select a subset of sensors which give the best probability of being correct given the data. This gives us a model from which we can determine whether sensors' readings fall out of a reasonable range for the sensor set. We apply our method to simulated data and actual environmental data collected in the forest.

Current sensor networks experience many faults that hamper the ability of scientists to draw significant inferences. We develop a method to systematically identify when these faults occur so that proper corrective action can be taken. We propose an adaptable modular framework that can utilize different modeling methods and approaches to identifying trustworthy sensors. We focus on using hierarchical Bayesian space-time (HBST) modeling to model the phenomenon of interest, and use maximum a posteriors selection to identify a set of trustworthy sensors. Compared to an analogous linear autoregressive system, we achieve excellent fault detection when the HBST model accurately represents the phenomenon.

To aid intelligence analysts in processing ambiguous data regarding nuclear terrorism threats, we develop a methodology that captures and accounts for the uncertainty in new information and incorporates prior beliefs on likely nuclear terrorist activity. This methodology can guide the analyst when making difficult decisions regarding what data are most critical to examine and what threats require greater attention. Our methodology is based on a Bayesian statistical approach that incorporates ambiguous cues to update prior beliefs of adversary activity. We characterize the general process of a nuclear terrorist attack on the United States and describe, using a simplified example, how this can be represented by an event tree. We then define hypothetical cues for the example and give notional strengths to each cue. We also perform sensitivity analysis and show how cue strengths can affect inference. The method can be used to help support decisions regarding resource allocation and interdiction.

No abstract

The following annotated bibliography was developed as part of the Geospatial Algorithm Verification and Validation (GSV) project for the Simulation, Algorithms and Modeling program of NA-22. Verification and Validation of geospatial image analysis algorithms covers a wide range of technologies. Papers in the bibliography are thus organized into the following five topic areas: Image processing and analysis, usability and validation of geospatial image analysis algorithms, image distance measures, scene modeling and image rendering, and transportation simulation models. Many other papers were studied during the course of the investigation including [1-17]. The annotations for these articles can be found in the paper "On the verification and validation of geospatial image analysis algorithms," [18]. Preface to version 2.0, 10 June 2011 The updated version of the GSV annotated bibliography contains several new citations. These citations were discussed in our weekly teleconferences, and were also used in the paper "Design of benchmark imagery for validating facility annotation algorithms," [19]. The citations include a survey of image annotation methods, an old but very relevant paper on image keys for identifying facilities, a paper discussing validation data for computer vision algorithms, and two papers on experimental design and validation methodologies for geospatial image analysis algorithms. For ease of review, these papers are grouped together and discussed in Section 6.