22-03-2012, 02:41 PM
Sequence-Based Localization in Wireless Sensor Networks
localization.pdf (Size: 324.82 KB / Downloads: 161)
INTRODUCTION
ACCURATE localization is an essential part of many
wireless sensor network applications. Over the years
many researchers have proposed many different solutions for
this problem ( [1], [2], [3], [4], [5], [6], [7], [8], [9], [10],
[11]). In these techniques, there is a tradeoff between the
accuracy of localization and the complexity of implementation.
For instance, least squares estimation techniques ( [1])
require accurate RF channel parameters such as the radio path
loss exponent; finger-printing based techniques (such as [8])
require extensive pre-configuration studies that depend on the
features of the localization space; other techniques require specialized
hardware ( [5]) or a complex configuration procedure
( [11]). On the other extreme, really simple techniques such
as computing centroid of nearby beacons ( [7]) provide low
accuracy. In this paper, we present a novel sequence-based
RF localization technique that is lightweight, works with any
hardware and provides accurate localization without requiring
accurate channel parameters or any pre-configuration.
LOCALIZATION USING LOCATION SEQUENCES
The procedure for localization of unknown nodes using
location sequences is as follows:
1) Determine all feasible location sequences in the localization
space and list them in a location sequence table.
2) Determine the location sequence of the unknown node
location using received signal strength (RSS) measurements
of localization packets exchanged between itself
and the reference nodes. The RSS based location
sequence will be a corrupted version of the original
location sequence.
3) Search in the location sequence table for the “nearest”
location sequence to the unknown node location
sequence. The centroid mapped to by that sequence is
the location estimate of the unknown node.
Feasible and Infeasible Sequences
As discussed previously, combinatorially, n reference nodes
produce O(nn) location sequences. But as shown in the
previous section, a localization space with n reference nodes
has only O(n4) distinct regions and consequently only O(n4)
feasible location sequences in the worst case. For given reference
node locations, the location sequence table includes all
feasible location sequences. All other sequences are infeasible.
The non-idealities of the RF channel could corrupt a feasible
location sequence either to another feasible sequence or an
infeasible sequence as illustrated in Figure 3. If the corrupted
sequence is infeasible, then it would be possible to detect the
corruption in the sequence, whereas, if the corrupted sequence
is feasible, corruption detection is not possible.
IV. LOCALIZATION SCENARIOS
In this section we illustrate two localization procedures for
two different scenarios that are determined by the localization
space size.
1) Entire localization space is within the radio range of the
unknown node: In this case, the location sequence table
remains constant for all locations of the unknown node
in the localization space. Therefore, the localization
procedure is as follows:
a) Pre-construct and store the location sequence table
using the locations of the reference nodes.
b) When the unknown node initiates the localization
process by broadcasting a localization packet, provide
the stored location sequence table along with
the RSS measurements from the reference nodes.
c) The unknown node determines its location sequence
using the RSS measurements and determines
its location by searching through the provided
location sequence table for the nearest feasible
location sequence.