Optimal placement of distributed sensors against moving targets

Abstract

This article addresses the problem of deploying a sparse network of sensors for surveillance of moving targets. The sensor networks of interest consist of sensors which perform independent binary detection on a target, and report detections to a central node for fusion. An optimization framework is developed for placement of sensors within a bounded search region, given sensor performance characteristics, prior information on anticipated target characteristics, and a distributed detection criteria. Individual sensor performance is represented parametrically as are priors on target dynamics. Several numerical examples are included that illustrate the utility of the optimization approach.

An Optimization Scheme for M2M-Based Patient Monitoring in Ubiquitous Healthcare Domain

“…While the coverage of a sensor network with immobile sensors has been extensively explored and studied by [12,13], researchers have recently studied the coverage of mobile sensor networks. Most of this work focuses on algorithms for repositioning of sensors in desired positions in order to enhance monitoring and tracking of the network coverage [14][15][16].…”

An efficient local improvement operator for the multi-objective wireless sensor network deployment problem

“…Examples of such approach are the works of Brass (2007), Cevher and M.Kaplan (2009), Lazos and Poovendran (2006), Manohar et al (2009), or Shrivastava et al (2009. In the second, the distribution function of the random node deployment can be optimized (for instance, a parametric function may be defined) so that the resulting network has the best possible performance statistics (Wettergren and Costa (2009)). …”

A probability density function approach to distributed sensors' path planning

“…It was shown in [4] that, when the number of sensors is very large, the probability of cooperative track detection is given by an integral function of the sensors' density, represented by a joint probability density function (PDF) of the sensors' positions in the region of interest. Recently, several authors addressed the fundamental problem of finding the sensors' ranges and positions that maximize the probability of track detection [1], [2], [6]. When the sensors are static, an approximately optimal sensors' distribution can be determined in the form of a S. Ferrari and G. Foderaro are with the Laboratory for Intelligent Systems and Control (LISC), Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC 27708-0005, {sferrari, greg.foderaro, andrew.tremblay}@duke.edu parameterized Gaussian mixture by computing the mixing proportions via sequential quadratic programming [6].…”

“…Recently, several authors addressed the fundamental problem of finding the sensors' ranges and positions that maximize the probability of track detection [1], [2], [6]. When the sensors are static, an approximately optimal sensors' distribution can be determined in the form of a S. Ferrari and G. Foderaro are with the Laboratory for Intelligent Systems and Control (LISC), Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC 27708-0005, {sferrari, greg.foderaro, andrew.tremblay}@duke.edu parameterized Gaussian mixture by computing the mixing proportions via sequential quadratic programming [6]. As shown in Section III, when the sensors are dynamic, the probability of track detection can be integrated over time, and optimized with respect to a time-varying Gaussian mixture using a finite number of collocation points.…”

“…One of the most effective approaches for escaping these local minima is to "fill" them, and to follow a new local path generated via random-walk algorithms. This paper shows how, after the Gaussian mixture is optimized by the approach presented in [6], the potential field approach can be used to efficiently compute collision-free paths for the individual sensors, such that their probability of detection is maximized, and their power required is minimized.…”