An unattended ground sensor (UGS) that attempts to perform target identification without providing some corresponding estimate of confidence level is of limited utility. In this context, a confidence level is a measure of probability that the detected vehicle is of a particular target class. Many identification methods attempt to match features of a detected vehicle to each of a set of target templates. Each template is formed empirically from features collected from vehicles known to be members of the particular target class. The nontarget class is inherent in this formulation and must be addressed in providing a confidence level. Often, it is difficult to adequately characterize the nontarget class empirically by feature collection, so assumptions must be made about the nontarget class. An analyst tasked with deciding how to use the confidence level of the classifier decision should have an accurate understanding of the meaning of the confidence level given. This paper compares several definitions of confidence level by considering the assumptions that are made in each, how these assumptions affect the meaning, and giving examples of implementing them in a practical acoustic UGS.

Target tracking and classification using passive acoustic signals is difficult at best as the signals are contaminated by wind noise, multi-path effects, road conditions, and are generally not deterministic. In addition, microphone characteristics, such as sensitivity, vary with the weather conditions. The problem is further compounded if there are multiple targets, especially if some are measured with higher signal-to-noise ratios (SNRs) than the others and they share spectral information. At the U. S. Army Research Laboratory we have conducted several field experiments with a convoy of two, three, four and five vehicles traveling on different road surfaces, namely gravel, asphalt, and dirt roads. The largest convoy is comprised of two tracked vehicles and three wheeled vehicles. Two of the wheeled vehicles are heavy trucks and one is a light vehicle. We used a super-resolution direction-of-arrival estimator, specifically the minimum variance distortionless response, to compute the bearings of the targets. In order to classify the targets, we modeled the acoustic signals emanated from the targets as a set of coupled harmonics, which are related to the engine-firing rate, and subsequently used a multivariate Gaussian classifier. Independent of the classifier, we find tracking of wheeled vehicles to be intermittent as the signals from vehicles with high SNR dominate the much quieter wheeled vehicles. We used several fusion techniques to combine tracking and classification results to improve final tracking and classification estimates. We will present the improvements (or losses) made in tracking and classification of all targets. Although improvements in the estimates for tracked vehicles are not noteworthy, significant improvements are seen in the case of wheeled vehicles. We will present the fusion algorithm used.

The acoustic emissions of a ground vehicle contain a wealth of information, which can be used for vehicle classification, e.g. in the battlefield. However, features that are extracted from the acoustic measurements are time-varying and contain a lot of uncertainties, especially when the acoustic measurements are obtained from multiple terrains, which makes the classification challenging. In this paper we present our study on the multi-category classification of ground vehicles based on the acoustic data of four environmental conditions. The goal is to design one classifier that can operate in all four terrains without a priori knowledge of a specific terrain. We first perform the data pre-processing (including elimination of redundant records, processing of data distortion and generation of prototypes), feature extraction, and uncertainty analysis. We then develop the Bayesian classifier, and type-1 (T1) and interval type-2 (T2) fuzzy logic rule-based classifiers (FLRBC). These classifiers have similar architectures, consist of four sub-systems each for one terrain, and have one probability model (Bayesian classifier) or one fuzzy logic rule (T1 and interval T2 FLRBCs) for each kind of vehicle on each terrain. They differ in the way that this common architecture is implemented. We also present the results of the experiments to evaluate the performance of all classifiers. Experimental results reveal that (1) both the T1 and interval T2 FLRBCs have better performance than the Bayesian classifier, and the interval T2 FLRBC has better performance than the T1 FLRBC; (2) each classifier has a smaller average but a slightly larger standard deviation of classification error rates when the majority voting-based temporal decision fusion is applied; and (3) when the majority voting-based temporal decision fusion is applied, both the T1 and interval T2 FLRBCs have better performance than the Bayesian classifier, and the interval T2 FLRBC has better performance than the T1 FLRBC.

Various sparse array configurations have been studied to improve spatial resolution for separating several closely spaced targets in tight formations using unattended acoustic arrays. To extend the array aperture, it is customary to employ sparse array configurations with uniform inter-array spacing wider than the half-wavelength intra-subarray spacing, hence achieving more accurate direction of arrival (DOA) estimates without using extra hardware. However, this larger inter-array positioning results in ambiguous DOA estimates. To resolve this ambiguity, sparse arrays with multiple invariance properties could be deployed. Alternatively, one can design regular or random sparse array configurations that provide frequency diversity, in which case every subarray is designed for a particular band of frequencies. These different configurations are investigated in this paper. Additionally, we present a Capon DOA algorithm that exploits the specific geometry of each array configuration. Simulation results are presented to study the pros and cons of different sparse configurations.

Prediction of acoustic transmission loss (TL), or the attenuation of sound pressure level (SPL) is a complex problem dependent on a variety of physical parameters. Prediction of the TL using a numeric parabolic equation (PE) method is often accepted as a method of providing accurate TL prediction, but the large computational time is a hinderance in applications requiring real-time situation awareness. In order to overcome these extreme computational requirements a neural network-based environmentally adaptive TL prediction method is proposed and developed in this paper. This method uses multiple back-propagation neural network (BPNN) predictors, each trained on specific environmental conditions, and then probabilistically combines the outputs of these predictors in a fusion center to obtain a final TL estimate. This method is implemented on a data set generated using a PE model for a wide range of geometric and environmental parameters. The results are then benchmarked against a single neural network-based prediction scheme.

The detection and localization of artillery guns on the battlefield is envisaged by means of acoustic aerial waves including sonic and infrasonic waves. The main objective of this work is to examine the different frequency ranges usable for the detection of an artillery gun on the battlefield and in particular to investigate the potential interest of the infrasound range. The main stages of this study have consisted of: - data acquisition of acoustic (including the infrasound range) signals of artillery guns and mortars, - modeling of the wave propagation in the atmosphere, - signal processing and evaluation of the performance of a system of networked sensors.

The MITRE Corporation has initiated a three-year internally-funded research program in netted sensors, the first-year effort focusing on vehicle detection for border monitoring. An important component is developing an understanding of the complex acoustic structure of vehicle noise to aid in netted sensor-based detection and classification. This presentation will discuss the design of a high-fidelity vehicle acoustic simulator to model the generation and transmission of acoustic energy from a moving vehicle to a collection of sensor nodes. Realistic spatially-dependent automobile sounds are generated from models of the engine cylinder firing rates, muffler and manifold resonances, and speed-dependent tire whine noise. Tire noise is the dominant noise source for vehicle speeds in excess of 30 miles per hour (MPH). As a result, we have developed detailed models that successfully predict the tire noise spectrum as a function of speed, road surface wave-number spectrum, tire geometry, and tire tread pattern. We have also included realistic descriptions of the spatial directivity patterns for the engine harmonics, muffler, and tire whine noise components. The acoustic waveforms are propagated to each sensor node using a simple phase-dispersive multi-path model. A brief description of the models and their corresponding outputs is provided.

This paper describes a robust detection algorithm implemented on a network of acoustic sensors. The sensors are severely constrained in both power and computational performance. A variety of techniques are employed to extract maximum detection range while minimizing false alarm rates under these constraints. These include automatic gain control, background estimation and adaptive thresholding, and collaboration among distributed sensors for false alarm mitigation. The resulting algorithm is both robust and sufficiently general to be applied in a variety of sensor domains. The algorithm was implemented and deployed on prototype hardware and operated in real time under realistic operational conditions.

This paper describes development and application of a high-fidelity, seismic/acoustic simulation capability for battlefield sensors. The purpose is to provide simulated sensor data so realistic that they cannot be distinguished by experts from actual field data. This emerging capability provides rapid, low-cost trade studies of unattended ground sensor network configurations, data processing and fusion strategies, and signatures emitted by prototype vehicles. There are three essential components to the modeling: (1) detailed mechanical signature models for vehicles and walkers, (2) high-resolution characterization of the subsurface and atmospheric environments, and (3) state-of-the-art seismic/acoustic models for propagating moving-vehicle signatures through realistic, complex environments. With regard to the first of these components, dynamic models of wheeled and tracked vehicles have been developed to generate ground force inputs to seismic propagation models. Vehicle models range from simple, 2D representations to highly detailed, 3D representations of entire linked-track suspension systems. Similarly detailed models of acoustic emissions from vehicle engines are under development. The propagation calculations for both the seismics and acoustics are based on finite-difference, time-domain (FDTD) methodologies capable of handling complex environmental features such as heterogeneous geologies, urban structures, surface vegetation, and dynamic atmospheric turbulence. Any number of dynamic sources and virtual sensors may be incorporated into the FDTD model. The computational demands of 3D FDTD simulation over tactical distances require massively parallel computers. Several example calculations of seismic/acoustic wave propagation through complex atmospheric and terrain environments are shown.

General Sensing Systems (GSS) has been developing seismic sensors for different security and military applications for the past several years. Research and development in this area does not have a single-value purpose as security and military applications are of a broad variety. Many of the requirements for seismic sensors are well known. Herein we describe additional requirements for seismic sensors that are not at the center of common attention and associated with high performance seismic sensors. We find that the hard issues related to "remote" deployment/installation methods can be solved, given the seismic sensor does not have the usual single-axis sensitivity, but sensitivity to arbitrary oriented impact/vibrations. Our results show that such a sensor can be designed, in particular based on electret materials. We report that traditional frequency response curve linearity is not always the appropriate goal. Such issues as useful signal frequency band and an interference immunity should be directly taken into account. In addition, the mechanical oscillator of the seismic sensor should have a very broad dynamic range about 120dB, or an adjustable sensitivity for use in various tactical applications. We find that increasing sensitivity is not so much needed as is reducing of the seismic sensor sensitivity threshold. The lower sensitivity threshold in higher target detection range can be obtained in low noise environmental conditions. We will also show that the attempt to design and manufacture a universal seismic sensor for every possible application seems unreasonable. In every respect it makes sense to design a seismic sensor set, which can fit and satisfy all plurality of the applications and multi objective requirements.

We describe a novel approach to target counting and speed estimation using the envelope of the seismic and acoustic signals. This approach is particularly well suited to low-cost, low-power sensors that lack the ability to determine target bearing and separation. We present methods for envelope peak detection, target detection, and speed estimation. Results of a field evaluation of one implementation of the algorithms are also presented.

In this paper we have demonstrated an unattended seismic wave sensor based on the optical fiber Bragg grating (FBG) sensing technology. The basic principle of seismic wave detection and the fiber Bragg grating sensing technique is introduced in brief. Our FBG seismic sensor system consists of the broadband light source, FBG sensors coupled with spring / mass configuration, 3-dB optical couplers, demodulator FBG gratings, signal detection and processing hardware. Systematic experiments were carried out in Fort Dix at New Jersey. Target source includes personnel, military HMMWV and wheeled truck. The FFT analysis shows that the frequency response of the seismic signal is 20-40 Hz. The comparison data also show that the FBG sensor is more sensitive than the conventional seismic sensor both in personnel and vehicle detection. The ultimate object of this imbedded fiber optic sensor system is to recognize the presence of troop or vehicle movement through their induced seismic activity as a helpful tool in battlefield monitoring and perimeter defense system.

The increasing need for unattended sensor networks drives individual sensor development, signal processing for network management, and communication technology. The application space is becoming more complex, with requirements for sensor networks in force protection; surveillance of large expanses of rugged terrain; and monitoring complex urban areas. Individual sensors exhibit excellent performance and include a wide variety of sensing modes, such as acoustic, electro-optical imaging, seismic, and radio frequency devices. These sensors continue to shrink with packaging, while applications continue to demand more of the sensor technology. Although single imaging arrays, which are available in spectral bands from the visible through the infrared, can be integrated into packages size as small as a cubic inch, the information from a single sensor is not sufficient to meet requirements for day/night, all-weather operation. This has driven the need for integration of multiple sensors into the compact packages intended for an individual sensor. A major step toward addressing the need for more effective sensor technology for unattended sensor networks is being taken through development of Vertically Integrated Sensor Array (VISA) technology. This technology, currently being developed for imaging sensors, builds multiple layers of signal processing at each pixel in the sensor array. Processing power is dramatically increased, allowing the integration of multiple sensors in small compact packages. This paper reviews the VISA approach to imaging sensors and describes applications for unattended sensors.

New generations of smart sensors must provide military and law enforcement agencies with reliable perceptual systems that are similar to human vision. The traditional approach cannot provide a reliable separation of an object from its background/clutter, while human vision unambiguously solves this problem. Vision is only a part of a system that converts visual information into knowledge structures. These structures drive the vision process, resolving ambiguity and uncertainty via feedback, and provide image understanding, which is an interpretation of visual information in terms of these knowledge models. These mechanisms provide a reliable recognition if the object is occluded or cannot be recognized as a whole. Biologically-inspired Network-Symbolic models convert image information into an "understandable" Network-Symbolic format, which is similar to relational knowledge models. Logic of visual scenes can be captured in the Network-Symbolic models and used for the disambiguation of visual information. Feature, symbol, and predicate are equivalent in the Network-Symbolic systems. A linking mechanism binds these features/symbols into coherent structures, and image can be interpreted by higher-level knowledge structures. View-based recognition is a hard problem for traditional algorithms that directly match a primary view of an object to a model. In Network-Symbolic Models, the derived structure, not the primary view, is a subject for unambiguous recognition.

The primary application considered in this paper is target detection in infra-red (IR) imagery for ground surveillance application. The problem is characterized by the need to rapidly search large areas, and ascertain the presence and location of threat objects. The challenge is in not only processing large images rapidly, but also reducing false alarm rates while achieving high levels of correct target detections. Performance is a function of many factors including the clutter type and density, the quality of the image and target signature, and the adequacy of the target data base. Although many types of target detection algorithms exist, we use a method based on quadratic correlation filters. The results of independent tests and evaluations by Night Vision Laboratory are also presented.

The DARPA Dynamic Optical Tags (DOTs) program has as its goal the development of a low cost, small, robust, persistent, low probability of intercept, 2-way tagging, tracking, and locating device that also supports error free data rates in excess of 100 kbps and can be interrogated at ranges up to and beyond 1Km. The program has selected several promising candidates for this device and is in the process of evaluating individually their performance against predetermined milestones to ascertain whether the technology is feasible and the program should continue for further development. In all cases the candidate devices operate as retro-reflecting optical modulators. Upon interrogation by a laser at the correct wavelength and with the correct code, the tags will proceed to modulate the return retro-reflection. While data for the candidate devices are not yet in hand, nevertheless this paper will provide an overview of the nature of the devices under investigation and speculate on how these devices could be employed for unattended ground sensor applications.

Human bodies are very good heat sources with peak emission wavelength of about 9?m. We use pyroelectric detectors that are differential in nature to detect human motion by their heat emissions. Coded Fresnel lens arrays create boundaries in space which helps to localize the human motion as well as classification. We design and implement a low-cost biometric tracking system using off-the-shelf components. We demonstrate tracking and classification using sensor clusters of dualelement pyroelectric detectors with coded Fresnel lens arrays.

The low-cost day and night imaging capability of uncooled infrared imagers significantly enhances the situational awareness capability of an unattended ground sensor. BAE Systems has leveraged its Standard Camera Core 500 product to develop an advanced imager, the MIM500TM, for use in unattended ground sensors and other military applications. Key improvements implemented in the MIM500TM include reduced operating power, pixel synchronization to an external clock, variable frame rate, a ruggedized mechanical design, a new reduced power standby mode, and electronic zoom,. This paper presents an overview of the MIM500TM design, it describes MIM500TM features that enhance the capability of unattended ground sensors, it discusses imaging performance data, and it provides an overview of current MIM500TM applications.

The design of an Unattended Ground Sensor (UGS) requires a tradeoff between cost and performance. For designs using a low cost IR microbolometer an array size of 160x120 pixels is a cost effective solution. However, this array size has limited resolving capability. Our goal is to make the best use of the available pixel information from this sensor. There are many reports describing super-resolution (SR) processing as a way to improve image resolution. The definition of SR adopted here is a process where a single high resolution image is created from a sequence of low resolution sub-pixel shifted images. The authors demonstrate the implementation of one SR algorithm from the literature and its benefits to UGS systems using both IR and visible imagery. We describe a software application where the analyst can input a low resolution image frame sequence to produce a high resolution output. The frame sequence can be of a globally shifted frame sequence, a static scene with moving objects or both.

The authors are developing a flexible sensor network with numerous potential uses. This paper presents one constructive application of the sensor integration platform and network. The theory of operation is as follows. Multiple sensing/processing nodes are scattered about a 30-­mile radius on land. Each node has the capability to handle up to a few different sensors based on user needs. The nodes operate autonomously while continuously acquiring data and process the information locally. The information deemed relevant to the system operator is uplinked to a base station at optimal intervals. Further data collection takes place at the base station where an operator can take appropriate action. Communication between the base station and nodes is based upon the emerging IEEE 802.16 standard. This enables broadband level information transfer at ranges capable of covering entire metropolitan areas. The application discussed in this paper is a homeland security monitoring system. This system would make use of a variety of nodes to collect data from surveillance cameras at border crossings and high value assets, water contamination sensors, weather sensors, and other sensors the user sees necessary. A technical description of the system architecture, its benefits, and limitations will be included. The utility of the 802.16 standard will also be incorporated in the paper.

This paper describes recent advances in the technology for, and implementation of, short-range optical communication links. The approach relies on molecular scattering of ultraviolet wavelengths by the atmosphere to achieve non-line-of-sight, omni-directional communication links. The same technology is also shown to be attractive for certain classes of line-of-sight links. A UV communication testbed implementation is described that is unique, employing research-grade semiconductor sources emitting in the solar-blind region of the UV spectrum, around 275nm. This paper extends previously reported field measurements to longer ranges and to a wider variety of application scenarios, including operation under tree canopy and operation in short-range quasi-line-of-sight links. Field measurements of atmospheric extinction at 275nm are reported and incorporated in a single-scatter propagation model to predict performance of line-of-sight links. Application of UV communication to foliage penetration uplinks is described, and performance is quantified through field measurements.

Historically, tactical sensor systems have transmitted limited amounts of data. Alert notifications, control signals, and status can be quickly transmitted using low rate data links such as 1200 bps. Increasingly, there is a desire to transmit more data through sensor systems. Signals may consist of E/O or IR images, acoustic or seismic signals, or near real-time target location information. Such capabilities are desired for Future Combat Systems, Objective Force Warrior, and the Objective Force. This paper addresses the ability of existing sensor systems to reliably provide timely transmission of large data files. Specifically, transmission of image files through sensor systems is analyzed. A theoretical analysis gives the probability of error-free image transmission, practical transmission distances, and the transmission time required. Experimental results that validate assumptions made in the theoretical analysis are given. Experimental results show that some existing sensor systems are fully capable of providing low latency, reliable image transmission.

The problems of sensor networking have generally been attacked using existing radio and networking technology and engineering. This paper describes several DARPA programs that are developing wireless, networking and power sources that have specific application to sensor and other low power network applications. Programs discussed include the DARPA Connectionless Networking (CN), Disruption Tolerant Networking (DTN) and Radio Isotope Micropower Sources (RIMS) programs.

Military forces conducting urban operations are in need of non-line-of-sight sensor technologies for enhanced situational awareness. Disposable sensors ought to be able to detect and track targets through walls and within rooms in a building and relay that information in real-time to the soldier. We have recently developed magnetic sensor nodes aimed towards low cost, small size, low power consumption, and wireless communication. The current design uses a three-axis thin-film magnetoresistive sensor for low bandwidth B-field monitoring of magnetic targets such as vehicles and weapons carried by personnel. These sensor nodes are battery operated and use IEEE 802.15.4 communication link for control and data transmission. Power consumption during signal acquisition and communication is approximately 300 mW per channel. We will present and discuss node array performance, future node development and sensor fusion concepts.

In an unattended implanted or mobile ground sensor scenario, the microcantilever platform is well suited: sensor power consumption has been demonstrated at the nanowatt level and, as a microelectromechanical system, the platform is inherently compact. In addition, the remarkable sensitivity, low cost, scalability, and versatility of microcantilever sensors make this technology among the most promising solutions for unattended ground sensing of chemical and biological agents, as well as explosives. The University of Nevada, Reno, and Nevada Nanotech Systems, Inc (NNTS) are currently developing a microcantilever-based detection system that will measure trace concentrations of explosives, toxic chemicals, and biological agents in air. A baseline sensor unit that includes the sensor array, electronics, power supply and air handling has been designed and preliminary demonstrations of the microcantilever platform have been conducted. The envisioned device would measure about two cubic inches, run on a small watch battery and cost a few hundred dollars. This first design is tailored to shipping container monitoring, but is a broadly applicable device for passive or active monitoring scenarios.

In order for the information provided by networks of unattended ground sensors (UGS) to be of use to the tactical planner, the location of each sensor must often be known. Sensor localization is typically achieved by careful hand emplacement, or facilitated by anchor nodes whose position is precisely known. Nova Engineering and Army Research Laboratory are currently designing a new sensor network architecture to meet the growing need for UGS networks that can self-localize using anchor nodes with imprecise prior location information. In this paper we present an overview of a prototype sensor network and an analysis of its capability. We also present new results on the effects of network deployment parameters on sensor localization error, such as the level of network connectivity and the number of GPS-enabled nodes on the network. We compare time of arrival (TOA) and time difference of arrival (TDOA) ranging algorithms for sensor localization, and we consider the benefits of including available direction of arrival (DOA) estimates. Selected scenarios of UGS deployments are provided for comparison.

Demands for miniaturized networked sensors that can be deployed in large quantities dictate that the packages be small and cost effective. In order to accomplish these objectives, system developers generally apply advanced packaging techniques to proven systems. A partnership of Nova Engineering and Tessera begins with a baseline of Nova's Unattended Ground Sensors (UGS) technology and utilizes Tessera's three-dimensional (3D) Chip-Scale Packaging (CSP), Multi-Chip Packaging (MCP), and System-in-Package (SIP) innovations to enable novel methods for fabricating compact, vertically integrated sensors utilizing digital, RF, and micro-electromechanical systems (MEMS) devices. These technologies, applied to a variety of sensors and integrated radio architectures, enable diverse multi-modal sensing networks with wireless communication capabilities. Sensors including imaging, accelerometers, acoustical, inertial measurement units, and gas and pressure sensors can be utilized. The greatest challenge to high density, multi-modal sensor networks is the ability to test each component prior to integration, commonly called Known Good Die (KGD) testing. In addition, the mix of multi-sourcing and high technology magnifies the challenge of testing at the die level. Utilizing Tessera proprietary CSP, MCP, and SIP interconnection methods enables fully testable, low profile stacking to create multi-modal sensor radios with high yield.

We introduce a truly modular, power-aware, distributed microsensor architecture, capable of seamlessly spanning performance metrics from point-optimized low-power to point-optimized high-power applications. This type of performance is often needed in unattended ground sensor applications such as acoustic sensing and tracking, where long periods of minimal sensing activity are intermixed with short periods of intense sensor processing. The system design and implementation of a microsensor platform based on this architecture are described with experimental results. We show that although building a modular power-aware system requires additional hardware components, it results in system capable of rapid physical hardware and software reconfiguration with module reuse for new applications, while achieving a significant decrease in overall system power.

The desire for persistent, long term surveillance and covertness places severe constraints on the power consumption of a sensor node. To achieve the desired endurance while minimizing the size of the node, it is imperative to use application-specific integrated circuits (ASICs) that deliver the required performance with maximal power efficiency while minimizing the amount of communication bandwidth needed. This paper reviews our ongoing effort to integrate several micropower devices for low-power wake-up detection, blind source separation and localization and pattern classification, and demonstrate the utility of the system in relevant surveillance applications. The capabilities of each module are presented in detail along with performance statistics measured during recent experiments.

In a previous paper we had described a novel concept on ultra-small, ultra-compact, unattended multi-phenomenological sensor systems for rapid deployment, with integrated classification-and-decision-information extraction capability from the sensed environment. Specifically, we had proposed placing such integrated capability on a 3-D Heterogeneous System on a Chip (HSoC). This paper amplifies two key aspects of that future sensor technology. These are the creation of inter-layer vias by high aspect ratio MPS (Macro Porous Silicon) process, and the adaptation of the TESH (Tori connected mESHes) network to bind the diverse leaf nodes on multiple layers of the 3-D structure. Interesting also is the inter-relationship between these two aspects. In particular, the issue of overcoming via failures, catastrophic as well as high-resistance failures, through the existence of alternative paths in the TESH network and corresponding routing strategies is discussed. A probabilistic model for via failures is proposed and the testing of the vias between the sensor layer and the adjacent processing layer is discussed.

Multiple target detection in sensor networks is a challenging problem since the signal captured by individual sensor nodes is normally a linear/nonlinear weighted mixture of geometrically diverse source signals. Independent component analysis (ICA) has been widely used to solve the source estimation problem, but most of the algorithms assume the number of sources is fixed and is equal to the number of observations, which generally is not the case in sensor networks. Even though several methods are put forward for source number estimation, traditional centralized schemes hinder their application in sensor networks due to extremely constrained resource and scalability issues. In this paper, a sequential source number estimation framework is developed, where we assume the sensor network has been self-organized into clusters. The determination of possible number of targets at a sensor is only based on its local observation and the estimation result received from its previous sensor. Therefore, raw data transmission is avoided and only small packets of partial estimation results are transmitted through the networks. Based on the local estimation generated within each cluster, a posterior probability fusion method based on the Bayes' theorem is derived. Experimental results show that using the sequential processing approach with probability fusion, the detection probability and reliability are improved over the centralized scheme while significantly reducing network traffic and conserving resources.

The cost of an unattended ground sensor system is based on two factors: the number of sensor nodes used and, the complexity of each sensor/communications node. The tracking accuracy of the sensor network is a trade off between the density of the network and the accuracy with which the sensor nodes can determine the position or bearing of the target. Assuming acoustic sensors, the errors reduce, primarily, to timing errors, within each of the sensor nodes. Therefore that understanding the timing errors within a network of acoustic nodes is a factor in determining system cost for a given level of information fidelity. This paper explores the error mechanisms within and without each of the sensor nodes thus identifying the critical sub systems where engineering effort would be most effectively directed.

The Micro Sensor System (MSS) is a layered sensor network with the goal of detecting dismounted threats approaching high value assets. A low power unattended ground sensor network is dependant on a network protocol for efficiency in order to minimize data transmissions after network establishment. The reduction of network 'chattiness' is a primary driver for minimizing power consumption and is a factor in establishing a low probability of detection and interception. The MSS has developed a unique protocol to meet these challenges. Unattended ground sensor systems are most likely dependant on batteries for power which due to size determines the ability of the sensor to be concealed after placement. To minimize power requirements, overcome size limitations, and maintain a low system cost the MSS utilizes advanced manufacturing processes know as Fluidic Self-Assembly and Chip Scale Packaging. The type of sensing element and the ability to sense various phenomenologies (particularly magnetic) at ranges greater than a few meters limits the effectiveness of a system. The MicroSensor System will overcome these limitations by deploying large numbers of low cost sensors, which is made possible by the advanced manufacturing process used in production of the sensors. The MSS program will provide unprecedented levels of real-time battlefield information which greatly enhances combat situational awareness when integrated with the existing Command, Control, Communications, Computers, Intelligence, Surveillance and Reconnaissance (C4ISR) infrastructure. This system will provide an important boost to realizing the information dominant, network-centric objective of Joint Vision 2020.

A decentralized version of particle swarm optimization called the distributed particle swarm optimization (DPSO) approach is formulated and applied to the generation of sensor network configurations or topologies so that the deleterious effects of hidden nodes and asymmetric links on the performance of wireless sensor networks are minimized. Three different topology generation schemes, COMPOW, Cone-Based and the DPSO--based schemes are examined using ns-2. Simulations are executed by varying the node density and traffic rates. Results contrasting heterogeneous vs. homogeneous power reveal that an important metric for a sensor network topology may involve consideration of hidden nodes and asymmetric links, and demonstrate the effect of spatial reuse on the potency of topology generators.

This study developed a distributed vehicle target detection and estimation capability using two algorithmic approaches designed to take advantage of the capabilities of networked sensor systems. The primary interest was on small, quiet vehicles, such as personally owned SUVs and light trucks. The first algorithm approach utilized arrayed sensor beamforming techniques. In addition, it demonstrated a capability to find locations of unknown roads by extending code developed by the Army Acoustic Center for Excellence at Picatinny Arsenal. The second approach utilized single (non-array) sensors and employed generalized correlation techniques. Modifications to both techniques were suggested that, if implemented, could yield robust methods for target classification and tracking using two different types of networked sensor systems.

Acoustic sensors have been the primary sensor of choice for many UGS network concepts. This is primarily due to their low cost, non line of sight performance and the fact that most targets of interest are noisy. This paper explores the benefits to be gained by attaching additional sensors to an acoustic sensor network to provide extra information. A methodology is described to assess the cost of acquiring a certain level of information and this is used to explore the context in which the sensor network is operated. It is demonstrated that the optimum choice of sensors is dependent on the target set and the information required from the network. The potential benefits of a 'plug and play' sensor suite are examined in the context of using this concept for targeting.

Increasingly, the U.S. is being drawn into non-traditional military operations in complex terrain, including villages and cities occupied by a combination of non-combatants and hostile forces. This trend will continue, as the world's urban population continues its significant growth. Of all the missions faced by U.S. military forces, combat and peacekeeping operations in urban terrain present the greatest challenges. In response, commanders will need to employ network-/execution-centric combined arms forces and Joint capabilities at the lowest tactical echelons and with minimal staffs. To do so, commanders will need information, especially from networked unattended ground sensors (UGS), to rapidly and effectively plan and rehearse urban operations, synchronize ground and air forces, command and control mixed assets, coordinate effects (from weapons and actions), and determine logistics support. Today, urban terrain operations’ tactics, techniques, and procedures (TTPs) are transforming from painstaking ground assaults with large numbers of forces and excessively destructive remote strikes, to more efficient Joint operations which identify, isolate, and assault nodes with precision weapons and smaller forces. Successful employment of such TTPs will support future command and control (C2) concepts; reduce force requirements, civilian casualties and collateral damage; and shorten mission times. These TTPs, future C2 concepts, and the complex urban environment will drive the networked UGS requirements for future operations. This paper will first address these emerging TTPs, future C2 concepts, and the complex urban environment; and identify associated networked UGS challenges. Next, the paper will discuss the recent uses of networked UGS in experiments and demonstrations. Finally, the paper will discuss the innovative employment of networked UGS in these experiments and demonstrations by Warfighters (e.g., for developing situational understanding, mission planning/rehearsal) and the many lessons learned about sensor information needs.

The MITRE Corporation has embarked on a three-year internally-funded research program in netted sensors with applications to border monitoring, situational awareness in support of combat identification, and urban warfare. The first-year effort emphasized a border monitoring application for dismounted personnel and vehicle surveillance. This paper will focus primarily on the Tier 1 acoustic-based vehicle classification component. We discuss the development and implementation of a robust linear-weighted classifier on a Mica2 Crossbow mote using feature extraction algorithms specifically developed by MITRE for mote-based processing applications. These include a block floating point Fast Fourier Transform (FFT) algorithm and an 8-band proportional bandwidth filter bank. Results of in-field testing are compared and contrasted with theoretically-derived performance bounds.

Potential terrorists/adversaries can exploit a wide range of airborne threats against civilian and military targets. Currently there is no effective, low-cost solution to robustly and reliably detect and identify low observable airborne vehicles such as small, low-flying aircraft or cruise missiles that might be carrying chemical, biological or even nuclear weapons in realistic environments. This paper describes the development of a forward-based fence that contains a multi-modal mix of various low cost, low power, netted sensors including unsophisticated radar, acoustic and optical (Infrared and visible) cameras to detect, track and discriminate such threats. Candidate target (Cessna, Beech Craft, crop duster, and cruise missile) signature phenomenologies are studied in detail through either theoretical, numerical simulation or field experiment. Assessments for all three modalities (Radar, acoustic and IR) indicate reasonable detectability and detection range. A multi-modal kinematic tracker is employed to predict the location, the speed and the heading of the target. Results from a notional, template based classification approach reveal reasonable discrimination between different aircraft tested in the field experiments.

Georgia Tech has initiated a research program into the issues surrounding the detection of covert personnel present in a wide variety of scenarios. These initial investigations have been focused on a detailed phenomenological analysis of human physiology with the subsequent identification and characterization of potential observables-particularly in the context of urban environments during this phase. A parallel effort focused on the characterization of the resulting humanspecific signatures and their dependency and variation on local environmental conditions. In addition, these studies established the basic requirements for the development of physics-based human signature models; a significant component of these models is the inclusion of environmental and physiological variables into the computations. This paper will present a review of the research program and preliminary results from several of the initial signature phenomenology tasks.

In this paper we examine the different functions performed by an acoustic-seismic unattended ground sensor (UGS) and how they contribute to the overall cost of the sensor and its implementation. A key point made is that the performance of the acoustic-seismic depends on target characteristics and environmental conditions and cannot easily be traded off against implementation costs. At the present time, sensor implementation costs are dominated by the radios. In general, it is best to use radios whose range is consistent with the detection capabilities of the sensors. Dense networks of sensors using short-range radios supporting distributed communication architecture will produce an unnecessary duplication of coverage and a much-increased cost. Different scenarios for the utilization of acoustic-seismic UGS are examined, and evaluated with respect to their implementation costs. The most cost-effective use of acoustic-seismic sensors is in those applications where they can use satellite communications or tap into an existing communications network.

Trafficking of illegal drugs on riverine and inland waterways continues to proliferate in South America. While there has been a successful joint effort to cut off overland and air trafficking routes, there exists a vast river network and Amazon region consisting of over 13,000 water miles that remains difficult to adequately monitor, increasing the likelihood of narcotics moving along this extensive river system. Hence, an effort is underway to provide remote unattended riverine detection in lieu of manned or attended detection measures.

A top concern in homeland security efforts is the lack of ability to monitor the thousands of miles of open border with our neighbors. It is not currently feasible to continually monitor the borders for illegal intrusions. The MicroSensor System (MSS) seeks to achieve a low-cost monitoring solution that can be efficiently deployed for border patrol applications. The modifications and issues regarding the unique requirements of this application will be discussed and presented. The MicroSensor System was developed by the Defense Microelectronics Activity (DMEA) for military applications, but border patrol applications, with their unique sensor requirements, demand careful adaptation and modification from the military application. Adaptation of the existing sensor design for border applications has been initiated. Coverage issues, communications needs, and other requirements need to be explored for the border patrol application. Currently, border patrol has a number of deficiencies that can be addressed with a microsensor network. First, a distributed networked sensor field could mitigate the porous border intruder detection problem. Second, a unified database needs to be available to identify aliens attempting to cross into the United States. This database needs to take unique characteristics (e.g. biometrics, fingerprints) recovered from a specialized field unit to reliably identify intruders. Finally, this sensor network needs to provide a communication ability to allow border patrol officers to have quick access to intrusion information as well as equipment tracking and voice communication. MSS already addresses the sensing portion of the solution, including detection of acoustic, infrared, magnetic, and seismic events. MSS also includes a low-power networking protocol to lengthen the battery life. In addition to current military requirements, MSS needs a solar panel solution to extend its battery life to 5 years, and an additional backbone communication link. Expanding the capabilities of MSS will go a long way to improving the security of the nation's porous borders.

A statistical model of Unattended Ground Sensor networks is developed and applied. The model assumes an UGS field comprised of clusters of UGS. Each cluster is formed by a random distribution of sensor nodes in an ad hoc network communicating with a central gateway. The model is exercised to evaluate different field parameters on model outputs such as track length, track accuracy, sensor area coverage, message throughput at choke points and so on. The model is joined with a simple model of distributed cluster management-a form of distributed processing-and the relative merits of this form of technology are assessed. The model is intended for initial UGS field design applications where it is desirable to examine very large input parameter sets or perform optimization calculations. The model has been implemented in a spreadsheet, but it is also suitable for a programmable calculator or solver program such as Matlab.

The urban environment poses a great many obstacles for the modern soldier, from complex buildings and streets to unknown or hidden combatants and non-combatants. To provide improved situational awareness and short range protection, a variety of sensors and sensor systems are under investigation and development. In order to provide timely information from small, low-cost sensor systems, ARL has been investigating the use of passive multi-modal sensors for the individual soldier. These sensors will combine several different sensing modalities, and combine the information from these sensors at the sensor level. This will improve the sensors ability to discriminate targets, reduce false alarms and minimize the amount of information required to be transmitted to the user. In addition, passive sensors are inherently lower power and more covert than active systems. This report will detail the initial accomplishments, and present early data on several sensing modalities under investigation.

Advances in technologies are providing opportunities to increase situational awareness in military operations. One application of these technologies is small disposable sensor systems that have the potential to enhance the war-fighter's lethality and survivability. Considering that the sensors must be disposable, cost constraints increase the complexity of solving the technical and operational challenges. This paper will address two areas of consideration when designing disposable sensor systems: technical and operational. Technical design considerations: Sensor communication networks are a hot area of development, with a multitude of standards and protocols to choose from. Miniaturization is providing a multitude of sensor modalities that can be considered for a disposable sensor. Decisions must be made between factors such as cost, size, and power requirements. Power management through hardware and software, coupled with more efficient batteries, is giving extended life to sensor systems. Environmental issues must be addressed in a disposable system. Operational design considerations: For systems monitored and managed directly by soldiers, one of the most important elements is the user interface, and its affect on overall system ease-of-use. System usefulness is also affected by the capability to autonomously monitor the area where sensors are deployed without a soldier present. The packaging of sensors will be affected by the current state of technology integrated into the sensors, as well as requirements for emplacement. These emplacement requirements and constraints will impact the operational effectiveness of the overall system.

This paper considers the problem of classifying ground vehicles using their acoustic signatures recorded by unattended passive acoustic sensors. Using these sensors, acoustic signatures of a wide variety of sources such as trucks, tanks, personnel, and airborne targets can be recorded. Additionally, interference sources such as wind noise and ambient noise are typically present. The proposed approach in this paper relies on the blind source separation of the recorded signatures of various sources. Two different frequency domain source separation methods have been employed to separate the vehicle signatures that overlap both spectrally and temporally. These methods rely on the frequency domain extension of the independent component analysis (ICA) method and a joint diagonalization of the time varying spectra. Spectral and temporal-dependent features are then extracted from the separated sources using a new feature extraction method and subsequently used for target classification using a three-layer neural network. The performance of the developed algorithms are demonstrated on a subset of a real acoustic signature database acquired from the US Army TACOM-ARDEC, Picatinny Arsenal, NJ.

The Aerial Canopy Sensor Delivery System (ACSDS) has been developed to meet the requirement of deep penetration and rapid, covert emplacement of persistent sensors and communications devices. The ACSDS allows deployment of small expendable packages (approx 4" diameter, up to 12" length, 3 pound payload), especially in areas that are otherwise impossible to reach or present unacceptable risk to the warfighter. Tests of the photovoltaic power generation, and rotochute air deceleration design have been accomplished. System power meets the design goal for average energy over a week. Flight tests show that the system is capable of controlled descent from a moving aircraft.