没有合适的资源?快使用搜索试试~ 我知道了~
range-free localization beyond connectivity
需积分: 0 1 下载量 123 浏览量
2009-11-18
10:42:21
上传
评论
收藏 4.88MB PDF 举报
温馨提示
试读
14页
range-free localization beyond connectivity , wireless sensor network
资源详情
资源评论
资源推荐
Achieving Range-Free Localization Beyond Connectivity
Ziguo Zhong
Department of Computer Science
University of Minnesota
zhong@cs.umn.edu
Tian He
Department of Computer Science
University of Minnesota
tianhe@cs.umn.edu
Abstract
Wireless sensor networks have been proposed for many
location-dependent applications. In such applications, the re-
quirement of low system cost prohibits many range-based meth-
ods for sensor node localization; on the other hand, range-free
localization that d e pend only on connectivity may underutilize
the proximity information embedded in ne ighborhood sensing.
In resp onse to the above limitatio ns, th is paper pr esents a novel
range-free appro ach to c apturing a relative distance between 1-
hop neighboring nodes from their neighborhood orderings that
serve as unique high-dimensional location signatures for nodes
in the network. With little additional overhead, the de sig n in
this p a per can b e conveniently applied as a transparent support-
ing layer for many state-of-the- art connectivity-based outdoor
sensor node loca liza tion solutions to achieve better positioning
accuracy than c onnectivity alone. We implemen ted our design
with three well-known localization algorithms and tested it in
two types of outdoor test-bed experiments: an 850- foot-long lin-
ear network composed of 54 MICAz motes, and a regular 2D
network covering an area of 10000 square feet with 49 motes.
System evaluation results show considerable performance gains
from the proposed design, which helps eliminate estimation am-
biguity with sub-hop resolution, and reduces localization errors
by as much as 3 5%. In addition, extensive simulations reveal an
interesting feature of robustness for our design under unevenly
distributed radio propagation path loss, and confirm its effec-
tiveness for large-scale outdoor networks.
Categories and Subject Descriptor s
C.2.4 [Computer Communication Networks]: Distributed
systems
General Terms
Algorithms, Design, Performa nce, Exper imentation
Keywords
Wireless Sensor Networks, Localization, Neighborhood Or-
dering, Regulated Signature Distance (RSD)
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation
on the first page. To copy otherwise, to republish, to post on servers or to redistribute
to lists, requires prior specific permission and/or a fee.
SenSys’09,
November 4–6, 2009, Berkeley, CA, USA.
Copyright 2009 ACM 978-1-60558-748-6 ...$5.00
1 Introduction
Wireless sensor networks (WSN) have been considered as a
promising tool for many location-de pendent applications [ 1, 2],
e.g., battlefield surveillance [3], e nvironment data collection [4],
event or human localization [5, 6]. In addition, some of the r out-
ing protocols [7, 8] and network management mechanisms pro-
posed for such networks are built on the assumption that geo-
graphic param eters of each sensor node are available. Although
sensor node localization plays an important r ole in all such sys-
tems, it is itself a challenging problem due to extremely limited
resources available at each low-cost and tiny sensor nod e .
Many excellent ideas have been proposed for node localiza-
tion in WSN. Based on whether accura te ranging is required,
there are basically two types of methods: (i) range-based loc al-
ization and (ii) range-free localizatio n. Range-based localiza-
tion could achieve good accuracy but is costly, requiring either
per-node ranging hardware [10, 12, 14, 16, 22] or careful sys-
tem calibration and environment pro filing [9, 11, 40], and thus is
not appr opriate for large-scale outdoor sensor networks. Range-
free approac hes localize nodes based on simple sensing, such
as wireless connectivity [26, 27, 29, 32, 33], anchor proxim-
ity [25, 28, 30], or localization events detec tion [36, 37]. Amo ng
these, connectivity-based solutions featu re a low overall system
cost, but by sacrificing localization a ccuracy.
Our work is motivated by the finding that localization by
means of mere connectivity may underutilize the proximity in-
formation available from n e ighborhood sensing. Although ra-
dio signal strength (RSS) is no t consid ered a good choice for
physical distance estimation in many scenario s because of un-
known radio path loss factors, mu lti-path effects, hardware dis-
crepancies, antenna orientation, and so forth [40, 41, 42], it
does provide some useful distance-related information beyond
indicating connectivity among neighboring nodes. Our experi-
mental study confirms th a t in outdoor open-air scenarios, the ra-
dio signal stren gth weakens approx imately monotonically with
the physical distance, especially from the v iewpoint of a sin-
gle no de, where RSS m ight provide heuristic inf ormation about
which neighboring nodes are closer and which are further.
Based on o ur empirical study, this p aper introduces a novel
range-free approach to extracting relative distance informa-
tion from neighborhood orderings which serve as unique high-
dimensional location signatures for sensor nodes in the net-
work. In stead of offering yet another new localization method,
the design described in this p aper can be conveniently ap-
plied as a transparent su pporting layer for many state-of-the-art
connectivity-based localization algorithms, provid ing a low-cost
but effective solutio n for greatly improving system accuracy.
We augmen te d three range-free localization algorithms -
MDS-MAP [26], DV-Hop [27], RPA [33], with our de sig n, and
evaluated the effectiveness of the proposed design in two types
of outdoor test-bed systems: an 850-fo ot-long linear network
with 54 MICAz motes, and a regular 2-dimensional network
covering an ar ea of 10000 square feet w ith 49 motes. System
evaluation showed noticeable performance g ains including elim-
inating estimation ambiguity and reducing localization errors b y
as much as 35%. In additio n, extensive simulation demonstrated
the effectiveness of our design for large-scale networks and re-
vealed an interesting feature of robustness to th e unkn own and
spatially unevenly distributed radio path loss.
The rest of the paper is organized as follows: Section 2 briefly
surveys related work. Sec tion 3 explains the motivation behind
our work with empirical data. Th e main design is introduced in
Section 4. Section 5 briefs three range-free protocols on which
we evaluated our work. Section 6 reports outdoor test-bed exper-
iments. Section 7 summaries results from extensive simulations.
Finally, Section 8 concludes the whole paper.
2 Related Work
Based on wheth er ranging is conducted at the resource-
constrained sensor nodes, most of the previous work about
node localization can be categorized into one of following two
classes: (i) range-based [ 9, 10, 11, 14, 12, 15, 16, 17, 18, 19, 20,
21, 22, 24], and (ii) range-free localization [25, 26, 27, 28, 30,
29, 31, 32, 34, 35, 36, 3 7, 38].
Range-based solutions tr y to estimate absolute distances
or angles among randomly deployed sensor nodes and then
apply triangulation or multilateration for location calculation.
Many range -based method s use techniques such as Time of
Arrival (ToA) [13], Time Difference of Arrival (TDoA) (e.g.,
Cricket [10], AHLos [11], TPS [12]) and Ang le of Arrival
(AOA) (e.g., A PS [22], SpinLoc [18]) to measure distance or an-
gles a mong nodes and anchors (also called beacons or reference
nodes with pre-known location information). Those methods
can be accurate but costly by adding per-node additional hard-
ware [10, 12, 14], requir ing intensive tuning [17] or not suitable
for large- scale systems due to their limited effective range [ 10].
Although some research has tried to utilize RSS (Receive Sig-
nal Str ength) with noise filtering for distance estimation or for
wireless fingerprin t match ing (e.g., Radar [ 9], wM D S [43], Spo-
tOn [45], Indoor GPS [46], Sequence [48], Ranking [49]), em-
pirical studies [39, 40, 41] have concluded that unless careful
calibration and environment profiling can be accomplished, RSS
is not a good choice for accu rate ranging.
Range-free methods have applied many smart ideas for pur-
suing a low-cost design. E a rly range-free solutions made use
of the proximity inf ormation to anchor no des. Typical exam-
ples are Centroid [25], APIT [28], Concave [35] and Self [30],
in which the high cost of anchors is supposed to be amor-
tized with a large number of low-cost ordinary senor nodes.
To achieve good accuracy, however, a high anchor density is
required, which is impractica l for large-scale systems. Con-
currently, wireless connectivity-based protocols such as DV-
Hop [27], M D S-MAP [26], RPA [33], Amorphous [32] and so
on, proposed using local neighborhood sensing to build h op-
based virtual distances for large-scale sensor ne twork localiza-
tion. In those sy stems, only a small number of anchors are
necessary for constru cting the globa l coordinates, which sig-
nificantly reduces the system cost. Recent work helps solve
the problems of “holes” [31, 34] and “complex shapes” [29],
contributing to connectivity-based solutions in practical irregu-
lar node deployment with obstacles. However, we found that
localization by means of connectivity alone does not make full
use of inform ation available from local neighborhood sensing.
Another important branch of range- free localization includ es
event-driven methods such as Lighthouse [36], Spotlight [37]
and etc., in which loc a lization events embedding temporal-
spatial relatio nships are generated and distributed across the net-
work area for determining nodes’ positions. Although event-
driven localiza tion p rovides tradeoffs between localization ac-
curacy and sy stem cost, generating localization events may not
be easy and convenient in some scenarios.
Instead of proposing yet another localization algorithm, this
paper presents the idea of regulated signature distance (RSD), as
a metric of the proximity among 1-hop neighboring nodes. Act-
ing as a transparent supporting layer, our design can effectively
improve the system accuracy of state-of-the-art connectivity-
based localization algor ithms with little extra cost.
3 Empirical Data as Motivation
This paper is motivated by our experim ental data showing
that in the ou tdoor environments,
• Network -wide m onotonic relationship between radio sig-
nal strength (RSS) a nd physical distance does not h old, but
• Per-node monotonic RSS-Distance relationship holds well,
i.e., any single node’s RSS sensing results fo r its neigh-
boring nodes can be used as an indic a tor for the relative
“near-far” relationship among neighbors.
In the following, we first explain results from a prelimin ary
test, and then pr ovide data obtained from large- scale outdoor
experiments for verification.
3.1 Preliminary Experiments
Figure 1 shows RSS sensing results from MICAz nodes in
several outdoor experiments conducted in two types of e nviron-
ment: grass land and parking lot. In the test, we placed 9 sender
nodes at different distances from a receiver node. Each sender
node broadcast 100 packets with 0dBm sending power, and the
receiver node recorded the RSS upon rece iving the packet. In
the grass-land scenario, we perform ed the test twice w ith two
different receiver nodes place d at the same location and without
moving or switchin g sender nodes (Grass Land Test1 and Grass
Land Test 2, respectively). In the parking lot scenario, id e ntical
sets of nodes were tested during day-time (Parking Lot Test 1)
and at night (Parking Lot Test 2). Tests were con ducted multiple
times, and results did not show significant changes in the overall
shapes of the curves shown in Figure 1.
0 10 20 30 40 50 60 70 80
−90
−85
−80
−75
−70
−65
−60
−55
Distance (in feet)
RSS (in dBm)
Grass Land Test 1
Grass Land Test 2
Parking Lot Test 1
Parking Lot Test 2
Figure 1. Experimental Results: RSS vs. Distance
From Figure 1, we can see that at the system level, using
absolute values of RSS for distance estimation is not reason-
able, because identical RSS values may correspond to different
distances. However, for each individual curve (i.e. from the
0 16 32 48 64 80 96 112 128 144
−100
−80
−60
−90dBm
↓
RSS vs. Physical Distance in Linear Network
Distance Between Node Pairs (in feet)
RSS (in dBm)
0 16 32 48 64 80 96 112
−100
−80
−60
−90dBm
↓
RSS vs. Physical Distance in Regular 2D Network
Distance Between Node Pairs (in feet)
RSS (in dBm)
(a) System Level View: RSS vs. Physical Distance
0 5 10 15 20 25 30 35 40 45 50 55
0
0.5
1
RSS Ordering vs. Distance Ordering in Linear Network
node ID
Similarity
0 5 10 15 20 25 30 35 40 45 50
0
0.5
1
RSS Ordering vs. Distance Ordering in Regular 2D Network
node ID
Similarity
(b) Each Node’s Point of View: Feature of RSS Monotonic At tenuation
Figure 2. Empirical Date from Large Scale Experiments
viewpoint of each sensing node), RSS values mostly decreased
monotonically with increasing distance, conveying information
about relative “near-far” relationships among 1-hop neighbors.
3.2 Large-Scale Experiments
We then conducted large-scale outdo or experimen ts with two
types of networks to verify the phenomena found in the prelim-
inary test. The first experiment was a linear network containing
54 MICAz nodes with a 1 6-foot intermediate distance between
adjacent nodes covering a 850-foot length along a road. In the
second experiment, we constructed a regular 2D network with a
7×7 grid-shaped layout including 49 nodes occupying an open-
air par king lot area of 10000 square feet. The setup of the large-
scale experiments will be further detailed in Section 6.
Figure 2 reports the empirical data obtained from the two
test-beds. Figure 2(a) plots the sensed RSS value for each pair
of nodes against the distance between the m, which verifies that
monotonic RSS-distance relationship doesn’t hold for the whole
network. In both the line ar network and the regular 2D network,
on one hand, RSS may vary dramatically f or identical distance.
For example, as shown in the right sub-figure of Figure 2(a),
RSS r anges from -60dBm to -90d Bm for a 16-foot distance in
the 2D network. On the other han d, a single RSS value may cor-
respond to a wide range of distances. For instance, as shown in
the left sub-figure, -90dBm could range from 32 fee t to 112 feet
in the linear network; even worse, -90 dBm RSS covers almost
all of the distance spectrum, i.e. from 16 feet to 112 feet, in the
2D network showing in the right sub-figure of Fig ure 2(a).
However, examining the data from the viewpoint of a single
node te lls a different story. For any node u
i
, we can obtain an
ordered node list, say A, by listing u
i
’s 1-hop neigh bors accord-
ing to their RSS values sensed at u
i
in decreasing order; and
another node list, say B, by ordering u
i
’s 1-hop neighbors by in-
creasing physical distance. Ideally, if the sensed RSS decreases
monotonically with increasing distance, A and B should be ide n-
tical. We d efine the similarity between two lists A and B as the
percentage of ac c ordant node pairs between them. For exam-
ple, let A = (u
1
,u
2
,u
3
) and B = (u
1
,u
3
,u
2
), then {u
1
, u
2
} is an
accordant node pair since in both A and B, node u
1
is ordered
ahead of u
2
; while { u
2
, u
3
} is not since their ordering gets re-
versed from A to B. We can see that if A and B are consistent
with their similarity close to 1, the monotonic feature holds.
Figure 2(b) illustrates the similarity results for all nodes in
two test-beds. We can see from the left sub-figure that in the
linear network, most of the n odes have a similarity close to
1 (the min imum, mean and maximum values of similarity are
0.86, 0.96 and 1, respec tively). It means th at in the linear net-
work, from single node’s po int of view, the RSS values for 1-hop
neighbors are approximately monotonic with the distance. This
finding still holds for the 2D regular network as shown in the
right sub-figure of Figure 2(b), wh e re the minimum, mean and
maximum similarities are 0.81 , 0.88 and 0.96, respectively.
Above experimen ts confirm that (i) RSS-distance relation-
ship does not hold at the system level, but (ii) the monotonic
feature approximately holds from the viewpoint of a single node.
3.3 Analysis and Discussion
This subsection discusses why the mo notonic RSS-distance
feature could hold from the viewpoint of a single node.
In addition to the physical distance between two nodes, there
are m a ny factors that affect RSS sensing results. Table 1 lists
some major aspects. We mar ked an aspect with a “
√
” if pre-
deploy en gineering efforts could possibly be applied to reduce
its impact, or a “×” if it would be hard or c ostly to address.
Table 1. Major Factors Affecting RSS Sensing
Types of Factors P
RF Transmit Parameters: Sending Power, Frequency, Modulation, Baud Rate ...
√
Antenna Issues: Transceiver Gain, Isotropic/directioinal, Orientation, He ight .. .
√
Random Noise: Interference, Nature Events, Mobile Effects, Electronic Pulse . ..
√
Propagation Path Loss: Terrain, Vegetation, Obstac le, Magnetic Field .. . ×
Node-level Sensing Discrepancy: LNA, IF, ADC Ref. Voltage, Ground Noise ... ×
At th e sender side, besides the sending power, th e carrier
frequency, modulation, baud rate a nd etc, determine the band-
width, center frequency and spectrum shape [47], which all af-
fect the RSS at the receiver side . Most of those parameters can
be co nfigured with small offset errors and maintain relatively
stable during the run time. Antenna issues such as isotropic
gain, orientation and etc, can also be carefully engineered in the
design phase. For transient random noise, traditional filter ing
methods are able to help reduce its impact. All of the above are
considered addressable without significant in-field calibra tion.
On the contrary, unpredictable e nvironmental factors are
much harder to handle. For example, radio path loss is unknown
and costly to profile in most cases because it is temporally dy-
namic and spatially unevenly distributed. Another difficult is-
sue is the sensing discrepancy among nodes. At the receiver
side, RSS sensing results are sensitive to small variance among
different nodes. For examp le , a tiny bias at the reference volt-
age of ADC or small variance of the LNA (low noise amplifier)
gain caused by different ground noise levels, may lead to dif-
ferent RSS values at two nodes even when their received signal
strengths are equivalent. Runtime sensing discr epancy among
剩余13页未读,继续阅读
jzkstc
- 粉丝: 2
- 资源: 8
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论0