Sensor placement

Over the past few decades many vibration and signal analysis techniques have been investigated, developed, and used to provide operational response information about mechanical power transmission systems for monitoring and diagnosis of components and their worn/faulted condition. The presence of multiple sources of excitation and forcing in a complex distributed mechanical structure (rotor transmission shafting, gearing, bearings, casing, and foundation) presents problems for algorithms that are designed to present a single feature associated with a single fault mechanism. The consequence of the mechanical structures dynamic interaction is not always clear but an understanding of it is critical for the successful development of effective integral signal p rocessing algorithms and automated reasoning components used in CBM systems. As part of a combined experimental-theoretical analysis effort, ARL is investigating mechanical fault evolution in damaged rotating components. A dynamics model of the system using component fault models was developed for response simulations. Comparison of experimentally measured and simulated results of system vibratory responses allow physical insights into vibratory measurement sensor placement and specification, dynamic system response to a fault, and the development of fault detection signal processing algorithms. In this paper, the dynamics modeling of the gearbox rotor/bearing-foundation system using the Finite Element method is outlined and it's relevance to diagnostics and prognostics is highlighted.

Maintaining insulative fat stores is vital for homeothermic marine mammals foraging in cold polar waters. To accomplish this, animals must balance acquisition and expenditure of energy. If this balance is shifted, body condition can decrease, challenging thermal homeostasis and further affecting energy balance. Prior studies of temperature regulation in sea lions have neither quantified basic all-inclusive heat flux values for animals swimming in cold water, nor determined whether they exhibit consistent spatial patterns of heat flux. Heat flux and skin temperature data were thus collected from four captive Steller sea lions using heat flux sensors (HFSs) with embedded thermistors. Optimal sensor placement was established using infrared thermography to locate the major areas of heat flux along the surface of the animals. Experiments were conducted on swimming animals in a large habitat tank with and without a drag harness, and on stationary animals in a temperature-and currentcontrolled swim flume. All heat flux measurements were corrected by a previously determined correction factor of 3.42 to account for insulative effects of the HFSs and attachment mechanism. Heat flux from shoulders and hips was consistently greater than from mid-trunk and axillary areas in both swimming and stationary animals, suggesting that certain areas of the body are preferentially used to offload excess heat. Mean heat flux for animals swimming with a drag harness was significantly greater than for unencumbered animals, indicating a likely increase in heat production beyond minimum heat loss. Thus, thermal stress does not appear to constitute significant costs for Steller sea lions swimming under conditions of increased drag at speeds of approximately 1 m/s in water temperatures of approximately 8.0 8C. D 2004 Elsevier B.V. All rights reserved.

The effectiveness of a sensor configuration for feedback flow control on the wake of a circular cylinder is investigated in both direct numerical simulation as well as in a water tunnel experiment. The research program is aimed at suppressing the von Kármán vortex street in the wake of a cylinder at a Reynolds number of 100. The design of sensor number and placement was based on data from a laminar two-dimensional simulation of the Navier-Stokes equations for the unforced condition. A low-dimensional proper orthogonal decomposition (POD) was applied to the vorticity calculated from the flow field and sensor placement was based on the intensity of the resulting spatial eigenfunctions. The numerically generated data was comprised of 70 snapshots taken over three cycles from the steady state regime. A linear stochastic estimator (LSE) was employed to map the velocity data to the temporal coefficients of the reduced order model. The capability of the sensor configuration to provide accurate estimates of the four low-dimensional states was validated experimentally in a water tunnel at a Reynolds number of 108. For the experimental wake, a sample of 200 particle image velocimetry (PIV) measurements was used. Results show that for experimental data, the root mean square estimation error of the estimates of the first two modes was within 6% of the desired values and for the next two modes was within 20% of the desired values. This level of error is acceptable for a moderately robust controller.

The positioning of nodes in a sensor network has an effect in its performance. In this paper, two predefined configurations are compared to the random distribution to gauge the magnitude of the effect on energy consumption of each type of sensor allocation. The experiments assume a flat, obstacle free, rectangular field, with Directed Diffusion used as routing protocol, and random different positions for the querying entity (sink) and the event location in the field. The results confirm that in an environment such as this, it is worth investing in the uniform positioning of sensors as they offer a significant performance gain.

We consider the role played by the sensor locations in the optimal performance of an array of acoustic vector sensors. First we derive an expression for the Cramér-Rao bound on the azimuth and elevation of a single far-field source for an arbitrary acoustic vector-sensor array in a homogeneous wholespace and show that it has a block diagonal structure, i.e., the source location parameters are uncoupled from the signal and noise strength parameters. We then derive a set of necessary and sufficient geometrical constraints for the two direction parameters, azimuth and elevation, to be uncoupled from each other. Ensuring that these parameters are uncoupled minimizes the bound and means they are the natural or "canonical" location parameters for the model. We argue that it provides a compelling array design criterion. We also consider a bound on the mean-square angular error and its asymptotic normalization, which are useful measures in three-dimensional bearing estimation problems. We derive an expression for this bound and discuss it in terms of the sensors' locations. We then show that our previously derived geometrical conditions are also sufficient to ensure that this bound is independent of azimuth. Finally, we extend those conditions to obtain a set of geometrical constraints that ensure the optimal performance is isotropic.

This paper presents a novel strategy in determining an optimal sensor placement scheme for environmental monitoring using Wireless Sensor Networks (WSN). This is accomplished by minimizing the variance of spatial analysis based on randomly chosen points representing the sensor locations. These points are assigned randomly generated measurements based on a specified distribution. Spatial analysis is employed using Geostatistical Analysis (classical variography and ordinary point kriging) and optimization occurs with Monte Carlo Analysis. A simple example of measuring mercury in soil is illustrated in finding the optimal sensor placement using WSNs. Studied variables include the number of sensor locations, variances, and Monte Carlo repetitions.

Activities of daily living are important for assessing changes in physical and behavioral profiles of the general population over time, particularly for the elderly and patients with chronic diseases. Although accelerometers have been used widely in wearable devices for activity classification, the positioning of the sensors and the selection of relevant features for different activity groups still pose significant research challenges. This paper investigates wearable sensor placement at different body positions and aims to provide a systematic framework that can answer the following questions: 1) What is the ideal sensor location for a given group of activities? and 2) Of the different time-frequency features that can be extracted from wearable accelerometers, which ones are the most relevant for discriminating different activity types?

This paper reviews Frechet sensitivity analysis for partial differential equations with variations in distributed parameters. The Frechet derivative provides a linear map between parametric variations and the linearized response of the solution. We propose a methodology based on representations of the Frechet derivative operator to find those variations that lead to the largest changes to the solution (the most significant variations). This includes an algorithm for computing these variations that only requires the action of the Frechet operator on a given direction (the Gateaux derivative) and its adjoint. This algorithm is applicable since it does not require an approximation of the entire Frechet operator, but only typical sensitivity analysis software for partial differential equations. The proposed methodology can be utilized to find worst case distributed disturbances and is thus applicable to uncertainty quantification and the optimal placement of sensors and actuators.

The key components in 2D dead reckoning navigation are the azimuth and the distance travelled. The devices used to obtain such information vary significantly with the application. In pedestrian navigation, the limitations on weight, size and ergonomics of the devices play important roles that make the classical inertial or GPS-aided inertial approach unsuitable. To overcome these constraints, other concepts have been investigated that integrate a digital magnetic compass and/or gyroscopes, bi-axial or tri-axial accelerometers, and altimeter with a single-frequency GPS receiver. The distance travelled is computed by merging the accelerometer signal with a physiological model whose parameters are calibrated on-line by GPS data.

Leaks are present to some extent in all water-distribution systems. This paper proposes a leakage localisation method based on the pressure measurements and pressure sensitivity analysis of nodes in a network. The sensitivity analysis using analytical tools is not a trivial job in a real network because of the huge non-explicit non-linear systems of equations that describe its dynamics. Simulations of the network in the presence and the absence of leakage may provide an approximation of this sensitivity. This matrix is binarised using a threshold independent of the node. The binary matrix is assumed as a signature matrix for leakages. However, there is a trade-off between the resolution of the leakage isolation procedure and the number of available pressure sensors. In order to maximise the isolability with a reasonable number of sensors, an optimal sensor placement methodology, based on genetic algorithms, is also proposed. These methodologies have been applied to the Barcelona Network using PICCOLO simulator. The sensor placement and the leakage detection and localisation methodologies are applied to several district management areas (DMA) in simulation and in reality.

This paper presents a technique for planning and controlling bevel-tip steerable needles towards a target location in 3-D anatomy under the guidance of partial, noisy sensor feedback. Our approach minimizes the probability that the needle intersects obstacles such as bones and sensitive organs by (1) explicitly taking into account motion uncertainty and sensor types, and (2) allowing for efficient optimization of sensor placement.We allow for needle trajectories of arbitrary curvature through duty-cycled spinning of the needle, which is conjectured to make a needle path small-time locally “trackable” [13]. This enables us to use LQG control to guide the needle along the path. For a given path and sensor placement, we show that a priori probability distributions of the needle state can be estimated in advance. Our approach then plans a set of candidate paths and sensor placements and selects the pair for which the estimated uncertainty is least likely to cause intersections with obstacles. We demonstrate the performance of our approach in a modeled prostate cancer treatment environment.

This paper reviews Fréchet sensitivity analysis for partial differential equations with variations in distributed parameters. The Fréchet derivative provides a linear map between parametric variations and the linearized response of the solution. We propose a methodology based on representations of the Fréchet derivative operator to find those variations that lead to the largest changes to the solution (the most significant variations). This includes an algorithm for computing these variations that only requires the action of the Fréchet operator on a given direction (the Gateaux derivative) and its adjoint. This algorithm is applicable since it does not require an approximation of the entire Fréchet operator, but only typical sensitivity analysis software for partial differential equations. The proposed methodology can be utilized to find worst case distributed disturbances and is thus applicable to uncertainty quantification and the optimal placement of sensors and actuators.

Over the past few decades many vibration and signal analysis techniques have been investigated, developed, and used to provide operational response information about mechanical power transmission systems for monitoring and diagnosis of components and their worn/faulted condition. The presence of multiple sources of excitation and forcing in a complex distributed mechanical structure (rotor transmission shafting, gearing, bearings, casing, and foundation)

Modelling of thermofluid systems is complex because of the coupling of thermal and hydraulic energies. Bond graph language is a suitable tool for modelling such nonlinear multienergy domain systems along with their control systems. Structural control properties of the system can be determined using the causal properties of bond graphs, without resorting to any formal computation. An ontology for classifying thermofluid process components is presented in this article, along with connection syntax and model validation algorithms. Then the theory for structural analysis is used to design sensor placements for observability and also for component fault detection and isolation. This study also deals with developing a bond graph-based model representation mechanism to perform structurallevel linking of submodels.The methodology developed is applied to two example systems. Simulation of process faults is used to validate the theoretically obtained fault signatures for a thermofluid system in the second example.

Activities of daily living are important for assessing changes in physical and behavioural profiles of the general population over time, particularly for the elderly and patients with chronic diseases. Although accelerometers are widely integrated with wearable sensors for activity classification, the positioning of the sensors and the selection of relevant features for different activity groups still pose interesting research challenges. This paper investigates wearable sensor placement at different body positions and aims to ...

The measured voltage signals picked up by the needle probe method can be interpreted by a numerical method so as to identify the magnetic material properties of the magnetic circuit of an electromagnetic device. However, when solving this electromagnetic inverse problem, the uncertainties in the numerical method give rise to recovery errors since the calculated needle signals in the forward problem are sensitive to these uncertainties. This paper proposes a stochastic Cramér-Rao bound method for determining the optimal sensor placement in the experimental setup. The numerical method is computationally time efficient where the geometrical parameters need to be provided. We apply the method for the non-destructive magnetic material characterization of an EI inductor where we ascertain the optimal experiment design. This design corresponds to the highest possible resolution that can be obtained when solving the inverse problem. Moreover, the presented results are validated by comparison with the exact material characteristics. The results show that the proposed methodology is independent of the values of the material parameter so that it can be applied before solving the inverse problem, i.e. as a priori estimation stage.

This paper discusses the current status of the development of the mobile aerial delivery system to be further employed in a variety of different applications. High accuracy of the developed system enables its use in precision troop resupply, precise sensors placement, urban warfare reconnaissance and other similar operations. This paper overviews the overall system architecture and components of the developed aero delivery system itself and then proceeds with describing the current status of integrating it with an advanced deployment platform, unmanned aerial system, to achieve mobility and autonomy of operations. The paper also discusses some other systems in development pursuing similar goals and reviews some novel applications that become possible with the developed aerial delivery system. 1 I.

This paper explores the opportunity of using tri-axial wireless accelerometers for supervised monitoring of sports movements. A motion analysis system for the upper extremities of lawn bowlers in particular is developed. Accelerometers are placed on parts of human body such as the chest to represent the shoulder movements, the back to capture the trunk motion, back of the hand, the wrist and one above the elbow, to capture arm movements. These sensors placement are carefully designed in order to avoid restricting bowler’s movements. Data is acquired from these sensors in soft-real time using virtual instrumentation; the acquired data is then conditioned and converted into required parameters for motion regeneration. A user interface was also created to facilitate in the acquisition of data, and broadcasting of commands to the wireless accelerometers. All motion regeneration in this paper deals with the motion of the human body segment in the X and Y direction, looking into the motio...

Daily life activities such as working and shopping may cause people to carry overloaded bags, frequently borne in an incorrect way (e.g. only on one shoulder, asymmetrically worn). When these activities alter the gait, back pain incidents can occur. Critical conditions can be monitored taking advantage from a wearable assistant, extracting contextual information by on-body acceleration signals. By acquiring data on trunk, limb and foot during gait, we are able to detect five walking tasks on loaded conditions: two-straps backpack carried on shoulders, backpack carried with a single strap on right and left shoulder, bag carried with the right and left hand. Seven subjects participated walking at self-selected speed on a treadmill carrying a load between 10-12% of their body weight. Subjects repeated each task for five times over three weeks. We classified the activities for a single user by use of KNN, naïve Bayes and SVM classifiers. KNN achieved the best recognition accuracy of 96.7% for day dependent classifier training. The sensors placement, which resulted to be different along consecutive days, affects performance evaluation: a +3 • rotation on the coronal plane decreases the accuracy to 76.0%.

The premise in low-authority control is that the actuators have limited authority and hence cannot signi cantly shift the eigenvalues of the system. As a result, the closed-loop eigenvalues can be well approximated analytically by perturbation theory. These analytical approximations may suf ce to predict the behavior of the closed-loop system in practice. We show that such approximations can be used to cast low-authority controller design problems for different objectives as convex optimization problems that can be solved ef ciently in practice by using recently developed interior-point methods. Also, we show that, by optimizing the l 1 norm of the feedback gains, we can arrive at sparse designs, i.e., designs in which only a small number of the control gains are nonzero. Thus, in effect, we can also solve actuator/sensor placement or controller architecture design problems. Examples are also given that demonstrate the effectiveness of the design method.

The key components in 2D dead reckoning navigation are the azimuth and the distance travelled. The devices used to obtain such information vary significantly with the application. In pedestrian navigation, the limitations on weight, size and ergonomics of the devices play important roles that make the classical inertial or GPS-aided inertial approach unsuitable. To overcome these constraints, other concepts have been investigated that integrate a digital magnetic compass and/or gyroscopes, bi-axial or tri-axial accelerometers, and altimeter with a single-frequency GPS receiver. The distance travelled is computed by merging the accelerometer signal with a physiological model whose parameters are calibrated on-line by GPS data.

Efficient camera placement is important in order to make sure the cost of a monitoring system is not higher than what it should be. This is also to ensure the maintenance of that system will not be complex and take longer time. Based on these issues, it has become an important requirement to optimize the number of the camera in camera placement system inside particular environment. This problem is based on the well-known Art Gallery Problem but most of previous works only proposed solution to this problem in 2D. We propose a method for finding the minimum number of cameras that can observe maximum space of 3D environment. In this method we assume that each of the cameras has limited field of view of 90o and only to be placed on the wall of the environment. Placement in 3D environment uses volume approach that takes frustum's volume and space's volume to calculate minimum number of camera.

Placement of sensors in water distribution networks helps timely detection of contamination and reduces risk to the population. Identifying the optimal locations of these sensors is important from an economic perspective and has been previously attempted using the theory of optimization. This work extends that formulation by considering uncertainty in the network and describes a stochastic programming method that is capable of determining the optimal sensor location while accounting for demand uncertainties. The problem is formulated as a two stage stochastic programming problem with recourse. The solution to the problem is achieved by using a newly proposed algorithm aimed at efficiently solving stochastic nonlinear programming problems. This makes the problem solution computationally tractable as compared to the traditional stochastic programming methods. The proposed formulation and solution methodology are tested on an example network to perform a comparative study with other formulations. The results show the importance of uncertainty consideration in decision making and highlight the advantages of the proposed stochastic programming approach.

In this paper, we study the Vapnik-Chervonenkis (VC)-dimension of set systems arising in 2D polygonal and 3D polyhedral configurations where a subset consists of all points visible from one camera. In the past, it has been shown that the VCdimension of planar visibility systems is bounded by 23 if the cameras are allowed to be anywhere inside a polygon without holes [1]. Here, we consider the case of exterior visibility, where the cameras lie on a constrained area outside the polygon and have to observe the entire boundary. We present results for the cases of cameras lying on a circle containing the polygon (VC-dimension=2) or lying outside the convex hull of a polygon (VC-dimension= 5). The main result of this paper concerns the 3D case: we prove that the VC-dimension is unbounded if the cameras lie on a sphere containing the polyhedron, hence the term exterior visibility.

This paper addresses the issue of monitoring physical spatial phenomena of interest using information collected by a resource-constrained network of mobile, wireless and noisy sensors that can take discrete measurements as they navigate through the environment. We first propose an efficient novel optimality criterion for designing a sampling strategy to find the most informative locations in taking future observations in order to minimize the uncertainty at all unobserved locations of interest. This solution is proven to be within bounds. The computational complexity of this proposition is shown to be practically feasible. We then prove that under a certain condition of monotonicity property the approximate entropy at resulting locations obtained by our proposed algorithm is within 1 − 1 e of the optimum, which is then utilized as a stopping criterion for the sampling algorithm. The criterion enables the prediction results to be within user-defined accuracies by controlling the number of mobile sensors. The effectiveness of the proposed method is illustrated using a pre-published data set.

This paper addresses the issue of monitoring physical spatial phenomena of interest utilizing the information collected by a network of mobile, wireless and noisy sensors that can take discrete measurements as they navigate through the environment. The spatial phenomenon is statistically modelled by a Gaussian Markov Random Field (GMRF) with hyperpa-rameters that are learnt as the measurements accumulate over time. In this context, the GMRF approximately represents the spatial field on an irregular lattice of triangulation by exploiting a stochastic partial differential equation (SPDE) approach, which benefits remarkably in computation due to the sparsity of the precision matrix. A technique of the one-step-ahead forecast is employed to predict the future measurements that are required to find the optimal sampling locations. It is shown that optimizing the sampling path problem with the logarithm of the determinant either of a covariance matrix using a GP model or of a precision matrix using a GMRF model for mobile robotic wireless sensor networks (MRWSNs) even by a greedy algorithm is impractical. This paper proposes an efficient novel optimality criterion for the adaptive sampling strategy to find the most informative locations in taking future observations that minimize the uncertainty at unobserved locations. The computational complexity of our proposed method is linear, which makes the MRWSN scalable and practically feasible. The effectiveness of the proposed approach is compared and demonstrated using a pre-published data set with appealing results.

A test protocol is defined for the purpose of measuring rectus femoris mechanomyographic (MMG) signals. The protocol is specified in terms of the following: measurement equipment, signal processing requirements, human postural requirements, test rig, sensor placement, sensor dermal fixation, and test procedure. Preliminary tests of the statistical nature of rectus femoris MMG signals were performed, and Gaussianity was evaluated by means of a two-sided Kolmogorov-Smirnov test. For all 100 MMG data sets obtained from the testing of two volunteers, the null hypothesis of Gaussianity was rejected at the 1%, 5%, and 10% significance levels. Most skewness values were found to be greater than 0.0, while all kurtosis values were found to be greater than 3.0. A statistical convergence analysis also performed on the same 100 MMG data sets suggested that 25 MMG acquisitions should prove sufficient to statistically characterize rectus femoris MMG. This conclusion is supported by the qualitative characteristics of the mean rectus femoris MMG power spectral densities obtained using 25 averages.

Activities of daily living are important for assessing changes in physical and behavioral profiles of the general population over time, particularly for the elderly and patients with chronic diseases. Although accelerometers have been used widely in wearable devices for activity classification, the positioning of the sensors and the selection of relevant features for different activity groups still pose significant research challenges. This paper investigates wearable sensor placement at different body positions and aims to provide a systematic framework that can answer the following questions: 1) What is the ideal sensor location for a given group of activities? and 2) Of the different time-frequency features that can be extracted from wearable accelerometers, which ones are the most relevant for discriminating different activity types?

For RSS-based methods a massive collection of training RSS samples is needed to calibrate the positioning system and to achieve a high positioning quality. The quality of these methods is directly related to the placement of the wireless sensors in the workspace and the Radio Map used to compute the user location. Traditionally deploying the reference points and building the Radio Map require human intervention and are extremely time-consuming. In this paper we aim to reduce these manual calibration efforts. We propose an automatic approach both to build a Radio Map in the given environment and to assess the best system calibration that fits the required positioning quality. The approach has been tested on the most used radio frequency-based technologies, i.e., IEEE 802.11 and Bluetooth.

Currently, the only high-quality, high-resolution temperature and precipitation data sets for the continental United States suitable for use on climatological time scales are for mean values. None yet exist that represent sequential monthly values over an extended historical ...

Much of the analysis of the potential to save energy and reduce emissions of greenhouse gases in the buildings sector has focused on the energy savings that can be achieved through incremental improvements in the efficiency of individual energy-using devices (motors, fans, pumps, boilers, chillers) but without changing the way in which they are put together as systems. However, much larger savings are possible through changes in building systems, and further, these savings can be achieved at much smaller incremental investment cost and sometimes at lower first cost. This paper will focus on the savings that can be achieved through a systems approach to buildings. The systems approach begins with a consideration of building shape and form and the specification of a high-performance envelope so as to minimize heating and cooling loads. This leads to a number of synergies that further reduce energy requirements, such as permitting cooler temperatures for distributing heat and warmer te...

We propose a novel framework for location detection with sensor networks, based on the theory of identifying codes. The key idea of this approach is to allow sensor coverage areas to overlap so that each resolvable position is covered by a unique set of sensors. In this setting, determining a sensor-placement with a minimum number of sensors is equivalent to constructing an optimal identifying code, an NP-complete problem in general. We thus propose and analyze new polynomial-time algorithms for generating irreducible (but not necessarily optimal) codes for arbitrary topologies. Our algorithms incorporate robustness properties that are critically needed in harsh environments. We further introduce distributed versions of these algorithms, allowing sensors to selforganize and determine a (robust) identifying code without any central coordination. Through analysis and simulation, we show that our algorithms produce nearly optimal solutions for a wide range of parameters. In addition, we demonstrate a tradeoff between system robustness and the number of active sensors (which is related to the expected lifetime of the system). Finally, we present experimental results, obtained on a small testbed, that demonstrate the feasibility of our approach.

A methodology is presented for designing cost-effective optimal sensor configurations for structural model updating and health monitoring purposes. The optimal sensor configuration is selected such that the resulting measured data are most informative about the condition of the structure. This selection is based on an information entropy measure of the uncertainty in the model parameter estimates obtained using a statistical system identification method. The methodology is developed for the uncertain excitation case encountered in practical applications for which data are to be taken either from ambient vibration tests or from other uncertain excitations such as earthquake and wind. Important issues related to robustness of the optimal sensor configuration to uncertainties in the structural model are addressed. The theoretical developments are illustrated by designing the optimal configuration for a simple 8-DOF chain-like model of a structure subjected to an unmeasured base excitat...

The positioning of nodes in a sensor network has an effect in its performance. In this paper, two predefined configurations are compared to the random distribution to gauge the magnitude of the effect on energy consumption of each type of sensor allocation. The experiments assume a flat, obstacle free, rectangular field, with Directed Diffusion used as routing protocol, and random different positions for the querying entity (sink) and the event location in the field. The results confirm that in an environment such as this, it is worth investing in the uniform positioning of sensors as they offer a significant performance gain.

Objectives: Advances in surface electromyography (sEMG) techniques provide a clear indication that refinement of electrode location relative to innervation zones (IZ) is required in order to optimise the accuracy, relevance and repeatability of the sEMG signals. The aim of this study was to identify the IZ for the sternocleidomastoid and anterior scalene muscles to provide guidelines for electrode positioning for future clinical and research applications.

We consider the problem of deploying sensors in a large water distribution network, in order to detect the malicious introduction of contaminants. We show that a large class of realistic objective functions -such as reduction of detection time and the population protected from consuming contaminated water -exhibit an important diminishing returns effect called submodularity. We exploit the submodularity of these objectives in order to design efficient placement algorithms with provable performance guarantees. Our algorithms do not rely on mixed integer programming, and scale well to networks of arbitrary size. The problem instances considered in our approach are orders of magnitude (a factor of 72) larger than the largest problems solved in the literature. We show how our method can be extended to multicriteria optimization, 1 selecting placements robust to sensor failures and optimizing minimax criteria. We provide extensive empirical evidence on the effectiveness of our method on two benchmark distribution networks, and an actual drinking water distribution system of greater than 21,000 nodes.

Most existing camera placement algorithms focus on coverage and/or visibility analysis, which ensures that the object of interest is visible in the camera's field of view (FOV). However, visibility, which is a fundamental requirement of object tracking, is insufficient for automated persistent surveillance. In such applications, a continuous consistently labeled trajectory of the same object should be maintained across different camera views. Therefore, a sufficient uniform overlap between the cameras' FOVs should be secured so that camera handoff can successfully and automatically be executed before the object of interest becomes untraceable or unidentifiable. In this paper, we propose sensor-planning methods that improve existing algorithms by adding handoff rate analysis. Observation measures are designed for various types of cameras so that the proposed sensor-planning algorithm is general and applicable to scenarios with different types of cameras. The proposed sensor-planning algorithm preserves necessary uniform overlapped FOVs between adjacent cameras for an optimal balance between coverage and handoff success rate. In addition, special considerations such as resolution and frontal-view requirements are addressed using two approaches: 1) direct constraint and 2) adaptive weights. The resulting camera placement is compared with a reference algorithm published by Erdem and Sclaroff. Significantly improved handoff success rates and frontal-view percentages are illustrated via experiments using indoor and outdoor floor plans of various scales.

On-shoe acceleration and orientation sensors have revealed as a potentially powerful means for capturing aspects of human gait. The placement of sensors however has been done intuitively and mostly without quantitative evaluation of sensor positioning. Based on recorded signals of the five placement options sole, heel, toe-cap, instep and ankle we built SVM classifiers using orientation-based features and evaluate their performance on three activity classes level walking, going upstairs and going downstairs. Finally we present an approach to a placement-invariant classification model and discuss the benefit for a bipedal sensing setup.

Activity recognition from on-body sensors is affected by sensor degradation, interconnections failures, and jitter in sensor placement and orientation. We investigate how this may be balanced by exploiting redundant sensors distributed on the body. We recognize activities by a meta-classifier that fuses the information of simple classifiers operating on individual sensors. We investigate the robustness to faults and sensor scalability which follows from classifier fusion. We compare a reference majority voting and a naive Bayesian fusion scheme. We validate this approach by recognizing a set of 10 activities carried out by workers in the quality assurance checkpoint of a car assembly line. Results show that classification accuracy greatly increases with additional sensors (50% with 1 sensor, 80% and 98% with 3 and 57 sensors), and that sensor fusion implicitly allows to compensate for typical faults up to high fault rates. These results highlight the benefit of large onbody sensor network rather than a minimum set of sensors for activity recognition and prompts further investigation.

The lack of proprioceptive feedback is a serious deficiency of current prosthetic control systems. The Osseo-Magnetic Link (OML) is a novel humeral or wrist rotation control system that could preserve proprioception. It utilizes a magnet implanted within the residual bone and sensors mounted in the prosthetic socket to detect magnetic field vectors and determine the bone's orientation. This allows the use of volitional bone rotation to control a prosthetic rotator. We evaluated the performance of the OML using a physical model of a transhumeral residual limb.

The main concern in current wireless sensor networks is maximizing the network lifetime. Research activities conducted in this field rely on the partial utilization of the sensor network resources. Nonetheless, most of these works neglect the coverage problem. In this paper, we investigate the relationship between the network lifetime and the coverage problem. We show that efficient sensor network deployment, can lead to optimal energy saving. To do so, we propose an efficient placement algorithm addressing the case where the monitored area density is equal to two.

Coverage is a fundamental task in sensor networks. We consider the minimum cost point coverage problem and formulate a binary integer linear programming model for effective sensor placement on a grid-structured sensor field when there are multiple types of sensors with varying sensing quality and price. The formulation is general and can be adapted to handle situations where sensing is perfect, imperfect or uncertain, and the coverage requirements are differentiated. Unfortunately, the new model suffers from the intractability of the binary integer programming formulations. We therefore suggest approximation algorithms and heuristics. Computational results indicate that the heuristic based on Lagrangean relaxation outperforms the others in terms of solution quality.

In many applications, one has to actively select among a set of expensive observations before making an informed decision. For example, in environmental monitoring, we want to select locations to measure in order to most effectively predict spatial phenomena. Often, we want to select observations which are robust against a number of possible objective functions. Examples include minimizing the maximum posterior variance in Gaussian Process regression, robust experimental design, and sensor placement for outbreak detection. In this paper, we present the Submodular Saturation algorithm, a simple and efficient algorithm with strong theoretical approximation guarantees for cases where the possible objective functions exhibit submodularity, an intuitive diminishing returns property. Moreover, we prove that better approximation algorithms do not exist unless NP-complete problems admit efficient algorithms. We show how our algorithm can be extended to handle complex cost functions (incorporating non-unit observation cost or communication and path costs). We also show how the algorithm can be used to near-optimally trade off expected-case (e.g., the Mean Square Prediction Error in Gaussian Process regression) and worst-case (e.g., maximum predictive variance) performance. We show that many important machine learning problems fit our robust submodular observation selection formalism, and provide extensive empirical evaluation on several real-world problems. For Gaussian Process regression, our algorithm compares favorably with state-of-the-art heuristics described in the geostatistics literature, while being simpler, faster and providing theoretical guarantees. For robust experimental design, our algorithm performs favorably compared to SDP-based algorithms.