LOCALE: Collaborative Localization Estimation for Sparse Mobile Sensor Networks

LOCALE: Collaborative Localization Estimation for Sparse Mobile Sensor Networks

Pei Zhang (Carnegie Mellon University, USA) and Margaret Martonosi (Princeton University, USA)
DOI: 10.4018/978-1-61520-655-1.ch020
OnDemand PDF Download:
$37.50

Abstract

Mobile devices, by their nature, are very personal devices. As the field of mobile application matures, applications are beginning to include location and other context aware services. In addition, current research is trending to more peer-to-peer capable systems. They will often be very sparse for all or part of their operation because of mobility. While some of these devices localizes with fixed location beacons or per-node GPS, these methods are not always possible due to many constraints. This chapter focuses on a robust statistical method in mobile networks to both determine the location of the device and provide an estimation of the accuracy. This method is provides seamless operation despite the local density of a mobile network, providing the application with a meaningful measure of location with accuracy. While this chapter only focuses on localization, the methods discussed here can be applied to provide other estimation based in-system measurements.
Chapter Preview
Top

1 Introduction

Mobile devices have become ubiquitous devices used in daily life. Applications used on these devices today utilize various user contexts, including location, in order to perform greater personal services. Currently, mobile systems relies on a system of infrastructure, which both limits their coverage and bandwidth. As these devices becomes more widespread, current research in mobile systems are trending to more distributed systems with peer-to-peer capabilities. For both technical and logistical reasons, such networks will often be very sparse for all or part of their operation, sometimes functioning more as disruption-tolerant networks (DTNs).

In these mobile systems, it is common to require the knowledge of a node's physical location. This information can be recorded and extensively used as sensed data (Juang, 2002), in routing decisions (Yu, 2001), or for policy decisions. However, it is a non-trivial task to obtain the location both accurately and efficiently, especially in a sparse network. There have been works in dense-and-fixed networks for localization (Langendoen, 2003). Unfortunately, most of these methods require reliable connections between localizing nodes. When applied to a sparsely mobile network, these methods are inefficient, and often ineffective. In sparse systems, two methods are mainly used for localization: GPS (Zhang, 2005), and beacon triangulation (Moses, 2002). In the ZebraNet deployments, GPS localization, while producing accuracy within 100 meters, used nearly 50% of total system energy. Beacon triangulation, on the other hand, requires a high density of beacons, which is too costly for many, if not most, applications.

One way to localize sparse mobile nodes is by using a Global Positioning System (GPS) on each node. This technique is widely used in mobile phones, vehicle tracking systems, and others. Unfortunately, many prerequisites have to be met for GPS to function properly. For example, the GPS antenna needs an unobstructed view of the sky, making it difficult for use indoors or in urban canyons. Furthermore, power consumption of the GPS greatly shortens the lifetime of sensor nodes, while considerably increasing the cost of each device.

To solve these problems, several mobile networks use radio location beacons as localization references. Nodes use a variety of methods to measure their distance from multiple fixed or mobile beacons with known locations in order to estimate their own locations (Moses, 2002). However, this method requires that beacons cover all areas of the network where localization is desired. Since mobile devices are free to wonder, this can translate into unacceptably high infrastructure costs in order to provide the needed coverage (in sparse areas) and needed bandwidth (in dense areas).

To reduce infrastructure requirements in dense networks, many collaborative methods have been developed for dense networks. Typically, nodes in these networks localize by collaboratively deducing network topology and using several anchor beacons to calculate absolute locations (Langendoen, 2003). If nodes are mobile, however, these methods become inefficient or ineffective due to increasing communication overhead needed for the collaboration. Furthermore, in sparse areas of the mobile networks, these kinds of naive collaboration becomes unworkable due to the low number of nodes within each node's communication range.

Key Terms in this Chapter

Collaboration Localization: Is a localization method designed to reduce infrastructure requirements in dense networks in which nodes localize by collaboratively deducing network topology and using several anchor beacons to calculate absolute locations.

Distributed Estimation: Is an estimation method that uses a system of multiple autonomous agents and a random vector, which the agents wish to estimate. In general, distributed estimation problems have numerous practical applications such as in sensor networks or group decision problems.

Mobile Node Localization: Is a method to find a mobile node’s physical location.

Mobile Networks: Refers to any system of transmitters and receivers that sends radio signals over the air, such as Wi-Fi local network, cellular network or satellite network. Mobile networks allow the devices in the network to be mobile.

Sparse Networks: Are networks in which the nodes are settled at widely spaced intervals. Sparse networks are the opposite of dense networks.

Probabilistic Distribution Combination: Is a method of combining probabilistic distributions, that is, distributions that identify either the probability of each value of a random variable or the probability of the value falling within a particular interval.

Sensor Networks: Are networks consisting of multiple detection stations called sensor nodes with a communications infrastructure intended to monitor and record conditions at diverse locations. Monitored parameters may include temperature, humidity, pressure, vibration, sound intensity and vital body functions.

Disruption-Tolerant Networks: Are networks developed so that temporary or intermittent communications problems, limitations or anomalies have the least possible adverse impact. These networks can sustain communications even in the face of broken links and long delays.

Complete Chapter List

Search this Book:
Reset