scispace - formally typeset

Proceedings ArticleDOI

Enhancing Positioning Accuracy through Direct Position Estimators Based on Hybrid RSS Data Fusion

26 Apr 2009-pp 1-5

TL;DR: It is suggested that typical median estimator must be replaced by maximum likelihood estimator (mode) to enhance the positioning accuracy in future hybrid localization systems.

AbstractIn this paper, localization based on Received Signal Strength (RSS) is investigated assuming a path loss log normal shadowing model. On the one hand, indirect RSS-based estimation schemes are investigated; these schemes are based on two steps of estimation: estimation of ranges from RSS and then estimation of position using weighted least square approximation. We show that the performances of this type of schemes depend on the used estimator in the first step. We suggest that typical median estimator must be replaced by maximum likelihood estimator (mode) to enhance the positioning accuracy. On the other hand, a new direct RSS-based estimation scheme of position is proposed; Monte Carlo simulations show that the new estimator performs better than indirect estimators and can be reliable in future hybrid localization systems.

Topics: Estimator (61%), RSS (54%)

Summary (2 min read)

Introduction

  • Nowadays, Location Based Services (LBSs) are more and more required by people and industries.
  • This is the scope of the FP7 WHERE project [2].
  • The proposed direct approach consists in the estimation of position directly from RSS measurements without going through ranges.
  • These different estimators are evaluated by Monte Carlo simulations and show that mode estimator is the best indirect estimator and that the new direct estimator performs better than all direct schemes.

II. LOG NORMAL SHADOWING PATH LOSS MODEL

  • The simple analysis often used in coexistence studies limits the propagation characteristics to the large scale of the signal at given distances .
  • In mathematical terms, the mean received power (around which there will still be shadowing and multipath) will vary with distance with an exponential law.
  • The measured loss varies about this mean according to a zero-mean Gaussian random variable, Xσsh , with standard deviation σsh.
  • For each environment or/and radio link, a characteristic value of each parameter, np and σsh, is used.
  • Taking this into account, a constant level of noise can result in ever increasing error when RSS is used to estimate distance; if RSS noise is sufficient that the authors cannot tell the difference between 1 and 1.5m, they also cannot tell the difference between 10m and 15m.

A. Estimation of range from RSS

  • Thus, this estimator may be practical when no information about shadowing is available.
  • Once the MS get this knowledge, the best estimator will be the mode which is the ML estimator.
  • To better evaluate the performances of these different estimators, the authors derived for each estimator its variance.

IV. PROPOSED RSS-BASED DIRECT ESTIMATOR

  • The mathematical formulation of the proposed direct estimation scheme is described.
  • The authors notice that it has an additional trivial solution at origin which hopefully can be easily eliminated if it comes out from the optimization algorithm.

V. SIMULATIONS RESULTS AND DISCUSSIONS

  • The authors evaluate the performances of the set of studied estimators described in section III and IV through Monte Carlo simulations.
  • The different steps of the simulation are the following: The Fig. 2 and Fig. 3 are obtained respectively for indoor and outdoor scenarios with the parameters described in Table III.
  • Moreover, these figures suggest that the new proposed direct estimator performs better than direct schemes in the two different cases (indoor and outdoor).
  • Thus, the authors believe that the direct RSS-based estimation scheme of MS’s position may enhances the positioning accuracy.

VI. CONCLUSION

  • The authors studied hybrid RSS-based localization estimators assuming a path loss log normal shadowing model.
  • The authors distinguished direct from indirect schemes.
  • Indirect estimation schemes consist in two steps: estimation of ranges from RSS using mean, median or mode estimators; and estimation of location using weighted least square approximation on previously estimated ranges.
  • Furthermore, a new direct scheme for location estimation from RSS is proposed and analyzed.
  • Next step will be to evaluate performance in more realistic scenarios and especially by using more realistic path loss model with adequate parameters.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

HAL Id: hal-00379638
https://hal.archives-ouvertes.fr/hal-00379638
Submitted on 29 Apr 2009
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Enhancing positioning accuracy through direct position
estimators based on hybrid RSS data fusion
Mohamed Laaraiedh, Stéphane Avrillon, Bernard Uguen
To cite this version:
Mohamed Laaraiedh, Stéphane Avrillon, Bernard Uguen. Enhancing positioning accuracy through
direct position estimators based on hybrid RSS data fusion. Vehicular Technology Conference, 2009.
VTC Spring 2009. IEEE 69th, Apr 2009, Barcelona, Spain. pp.1 - 5. �hal-00379638�

Enhancing positioning accuracy through direct
position estimators based on hybrid RSS data fusion
M.Laaraiedh, S.Avrillon, B.Uguen
IETR, University of Rennes 1
{mohamed.laaraiedh, bernard.uguen, stephane.avrillon}@univ-rennes1.fr
Abstract In this paper, localization based on Received
Signal Strength (RSS) is investigated assuming a path loss log
normal shadowing model. On the one hand, indirect RSS-based
estimation schemes are investigated; these schemes are based on
two steps of estimation: estimation of ranges from RSS and then
estimation of position using weighted least square approximation.
We show that the performances of this type of schemes depend on
the used estimator in the first step. We suggest that typical median
estimator must be replaced by maximum likelihood estimator
(mode) to enhance the positioning accuracy. On the other hand, a
new direct RSS-based estimation scheme of position is proposed;
Monte Carlo simulations show that the new estimator performs
better than indirect estimators and can be reliable in future
hybrid localization systems.
Index Terms Localization, RSS, Indirect vs Direct Location
Estimation, Weighted least square, Hybrid Data Fusion, 4G
networks, ranging, Path Loss, Log Normal Shadowing.
I. INTRODUCTION
Nowadays, Location Based Services (LBSs) are more
and more required by people and industries. Security is the
main motivation for civilian mobile position location whose
implementation is nowadays mandatory for the emergency
calls. Besides security, the second leading application for
wireless localization is intelligent transportation systems
(ITSs). Personal tracking, navigation assistance and position-
dependent billing are also new LBSs in expansion [1].
Furthermore, the location information is not only valuable
for itself to provide new services but also to improve cellular
communication systems at various levels. This is the scope
of the FP7 WHERE project [2].
Location methods based on Received Signal Strength (RSS)
have an important advantage compared with others methods
since RSS is usually available whatever is the Radio Access
Network (RAN) [3]. Nevertheless, the precision and accuracy
of RSS is different from one RAN to another. The challenge
here is to merge hybrid RSSs characterized by different
accuracies and coming from different systems in order to
enhance the position accuracy. In the following, hybrid
RSS fusion relates to an algorithm which make use of RSS
observables coming from different RANs (Cellular, WLAN,
UWB, etc). This is the typical case in 4G networks where
nodes with different technological platforms are integrated
and in which the MS may be connected conjointly to cellular
Base Station (BS) and wireless Access Point (AP) [4].
Historically, RSS can be used in either fingerprinting or
lateration. Fingerprinting with RSS refers to the type of
algorithms that first collect RSS fingerprints of a scene and
then estimate the location of the MS by matching on-line
measurements with the closest location fingerprints [5]. RSS
lateration consists in estimating the ranges from collected
RSSs assuming a path loss model and then computing
position using these different estimated ranges. Generally, to
estimate range from RSS the median estimator is used [6],
[7], [8], [9]. This estimator do not require the knowledge of
shadowing which affects the RSS measurements, and it is
useful when no information about shadowing is available.
Nevertheless, in the case of a non Gaussian distribution, this
estimator performs worse than the Maximum Likelihood
estimator (ML) given by the mode of the distribution.
In the present study, we distinguish the indirect from
the direct schemes of positioning with RSS. The indirect
schemes are commonly used in previous works and consists
in positioning using ranges previously estimated from RSS
measurements assuming a path loss model. The proposed
direct approach consists in the estimation of position directly
from RSS measurements without going through ranges.
Assuming a log normal shadowing model for path loss,
three indirect estimators (mean, median and mode) are firstly
investigated; Then, the new direct estimation scheme is
proposed. These different estimators are evaluated by Monte
Carlo simulations and show that mode estimator is the best
indirect estimator and that the new direct estimator performs
better than all direct schemes.
The rest of the paper is organized as follows. Section II
investigates the log normal shadowing model and presents the
different radio propagation parameters which may affect the
positioning accuracy. Section III presents the three indirect
estimation schemes. Then, section IV proposes the new direct
estimation scheme and its mathematical formulation. In section
V, the performances of each estimator are evaluated and dis-
cussed using Monte Carlo simulations. Finally, our concluding
remarks are given in section VI.
In order to simplify the lecture of this paper, a list of
abbreviations and symbols that are used in the paper is given
in Table I.

AN Anchor Node
LS Least Square
ML Maximum Likelihood
MS Mobile Station
RSS Received Signal Strength
d Distance between transmitter and receiver (m)
d
0
Reference distance generally equal to 1 meter
L Pathloss at distance d (dB)
L
0
Pathloss at distance d
0
(dB)
n
p
Pathloss exponent
λ Wavelength (m)
σ
sh
Standard Deviation of shadowing (dB)
x =(x, y) Coordinates of the MS
x
k
=(x
k
,y
k
) Coordinates of the k
th
AN
l Length of the simulated area
N
Trial
Number of Trials in Monte Carlo simulations
TABLE I: List of different used abbreviations and symbols.
table
II. LOG NORMAL SHADOWING PATH LOSS MODEL
The simple analysis often used i n coexistence studies limits
the propagation characteristics to the large scale of the signal
at given distances (pathloss). In mathematical terms, the mean
received power (around which there will still be shadowing
and multipath) will vary with distance with an exponential
law. The total pathloss at a distance, d, will then be L, often
modelled as [10]:
L = L
0
+10n
p
log(
d
d
0
) (1)
d
0
, d, n
p
and L
0
are defined in Table I. L
0
is given by:
L
0
= 20 log(
4πd
0
λ
) (2)
In fact this expression of L represent only the mean loss of
the power. The measured loss varies about this mean according
to a zero-mean Gaussian random variable, X
σ
sh
, with standard
deviation σ
sh
. Shadowing is caused by obstacles between the
transmitter and receiver that attenuate signal power through
absorption, reflection, scattering, and diffraction. The complete
path loss equation expressed in dB is then given by:
L = L
0
+10n
p
log(
d
d
0
)+X
σ
sh
(3)
This model can be used for both indoor and outdoor
environments. For each environment or/and radio link, a char-
acteristic value of each parameter, n
p
and σ
sh
, is used. These
values can be determined by calibration via measurement
companions. Furthermore, the frequency and the bandwidth
affect these parameters. The most common values of n
p
are
shown by Table II for different types of environments.
Type of environment Path loss exponent n
p
Free Space 2
Urban area cellular radio 2.7 to 3.5
Shadowed urban cellular radio 3 to 5
InbuildingLOS 1.6to1.8
Obstructed in building 4 to 6
Obstructed in factory 2 to 3
TABLE II: Path Loss Exponent for different environments [4].
table
The log normal shadowing model is very interesting for
localization because it defines a linear relation between RSS
and the logarithm of the distance between MS and AN.
Nevertheless, the precision of estimated distance decreases
as the separation between MS and AN increases. As a rule
of thumb, if n
p
=2then RSS drops by 6 dB every time
distance doubles. This sub-linear attenuation rate means that
the difference in RSS between 1 m and 2 m is similar to the
difference between 10 m and 20 m: exactly 6 dB (Fig. 1).
Taking this into account, a constant level of noise can result in
ever increasing error when RSS is used to estimate distance;
if RSS noise is sufficient that we cannot tell the difference
between 1 and 1.5 m, we also cannot tell the difference
between 10 m and 15 m. As shown in Fig. 1, changes in RSS
due to distance become small relative to noise, even if the
level of noise remains the same over distance [11].
Fig. 1: Variation of path loss with respect to distance using Log
Normal Shadowing model: Error increases over distance depending
on both noise and attenuation rate. As the path loss flattens out,
differences in RSS become small relative to noise level.
figure
III. RSS-
BASED INDIRECT ESTIMATORS
In this section, we investigate the indirect RSS-based po-
sitioning schemes which consist in two steps: estimation of
ranges from RSS observables and then estimation of position
using weighted LS approximation on the previously estimated
ranges
A. Estimation of range from RSS
Let’s consider the log normal shadowing described by the
equation (3) as the used path loss model where we assume
that the shadowing term X
σ
sh
is zero-mean Gaussian :
X
σ
sh
∼N(0
2
sh
) (4)
From (3) and (4) we derive the fact that the distance d follows
a Log-Normal distribution :
p
d
(d, L)=
1
2πdS
e
(ln d M )
2
2S
2
(5)

where
S =
σ
sh
ln 10
10n
p
(6)
M =
(L L
0
)ln10
10n
p
+lnd
0
(7)
As d follows a Log-Normal distribution, the mean, median
and mode of estimated distance
ˆ
d are given respectively by
[12]:
ˆ
d
LS
= e
M+
S
2
2
(8)
ˆ
d
median
= e
M
(9)
ˆ
d
ML
= e
MS
2
(10)
From equations (8) to (10), one can notice that the only
estimator that does not consider the knowledge of shadowing,
given by the term S, is the median. Thus, this estimator may be
practical when no information about shadowing is available.
Once the MS get this knowledge, the best estimator will be
the mode which is the ML estimator. The mean estimator is
not a good choice as it over estimates the distance, and it is
very inaccurate especially for strong values of S.
To better evaluate the performances of these different esti-
mators, we derived for each estimator its variance. We obtained
the estimated variances of mean, median, and mode estimators
of distance are, respectively, given by:
ˆσ
2
LS
=
ˆ
d
2
LS
e
2S
2
(e
S
2
1) = e
2M+3S
2
(e
S
2
1) (11)
ˆσ
2
median
=
ˆ
d
2
median
e
S
2
(e
S
2
1) = e
2M+S
2
(e
S
2
1) (12)
ˆσ
2
ML
=
ˆ
d
2
ML
(1 e
S
2
)=e
2M2S
2
(1 e
S
2
) (13)
B. Weighted LS estimation of position
Once the MS gets the necessary amount of RSS observables
(3 at least in 2D scenario), it can perform the first step by
estimating the different ranges (
ˆ
d
k
)
k=1,..,K
with respect to the
K discovered AN in the scene. These ranges can be estimated
using one of the three estimators given by (8), (9) or (10).
Thus, we obtain the system :
(x x
1
)
2
+(y y
1
)
2
=
ˆ
d
2
1
...
(x x
K
)
2
+(y y
K
)
2
=
ˆ
d
2
K
(14)
Subtracting the first one (k =1)from others equations of
(14) results in
2
x
2
x
1
y
2
y
1
... ...
x
K
x
1
y
K
y
1
x
y
=
h
2
+
ˆ
d
2
1
ˆ
d
2
2
...
h
K
+
ˆ
d
2
1
ˆ
d
2
K
(15)
where h
k
= x
2
k
x
2
1
+ y
2
k
y
2
1
for k in (2, .., K).
The least square solution is then given by [6]:
x =
1
2
(A
T
A)
1
A
T
h (16)
where
A =
x
2
x
1
y
2
y
1
... ...
x
K
x
1
y
K
y
1
, x =
x
y
(17)
h =
h
2
+
ˆ
d
2
1
ˆ
d
2
2
...
h
K
+
ˆ
d
2
1
ˆ
d
2
K
(18)
In order to enhance the performances of LS regression,
we introduce the matrix of covariance of estimated ranges.
Three covariance matrices are then defined depending on used
ranges estimator. For the mean, median and mode estimator,
respetively, this covariance matrix is given by:
R
LS
= diag((ˆσ
2
LS,k
)
k=2,..,K
) (19)
R
median
= diag((ˆσ
2
median,k
)
k=2,..,K
) (20)
R
ML
= diag((ˆσ
2
ML,k
)
k=2,..,K
) (21)
The weighted least square solution is then given by [6]:
x =
1
2
(A
T
R
1
A)
1
A
T
R
1
h (22)
where R can be R
LS
, R
median
,orR
ML
.
IV. P
ROPOSED RSS-BASED DIRECT ESTIMATOR
In this fourth section, the mathematical formulation of the
proposed direct estimation scheme is described. To proceed,
let’s assume that the MS is connected to K ANs. For each
link k, the distribution of d
k
= x x
k
is given by equation
(5):
p
k
(xx
k
,L
k
)=
1
2πx x
k
S
k
e
(ln x x
k
−M
k
)
2
2S
2
k
=
1
2πd
k
S
k
e
(ln d
k
M
k
)
2
2S
2
k
(23)
In order to simplify the study, we assume the independence
of the K random variables (p
d
k
)
k=1,..K
. Hence, the conjoint
probability density function of these K random variables is
given by :
p
1...K
(d
1
...d
K
,L
1
...L
K
)=
K
k=1
1
2πd
k
S
k
e
(ln d
k
M
k
)
2
2S
2
k
(24)
Let’s introduce F (x)=ln(p
1...K
(d
1
...d
K
,L
1
...L
K
)). Thus
F is given by:
F (x)=
K
k=1
(ln(
2πd
k
S
k
)+
(ln d
k
M
k
)
2
2S
2
k
) (25)
The proposed new ML estimator is then defined by:
ˆ
x = min
x
F (x) (26)

Developing the expression of F leads to:
ˆ
x = min
x
F (x) = min
x
K
k=1
(ln x x
k
−(M
k
S
2
k
))
2
2S
2
k
(27)
To minimize F , we derived its gradient F . It can be
readily shown then that the proposed ML estimator follows
the implicit relation given by:
F (
ˆ
x)=
K
k=1
1
S
2
k
((M
k
S
2
k
) ln
ˆ
x x
k
)
ˆ
x x
k
ˆ
x x
k
ˆ
x x
k
= 0
(28)
We remark that this function F is ill-conditioned when
(M
k
S
2
k
) is in [0, 1]. In that case, the ML estimator may be
given by:
F (
ˆ
x).
ˆ
x = 0 (29)
This functional (29) share with the previous one (28) the
targeted position. However, we notice that it has an additional
trivial solution at origin which hopefully can be easily elimi-
nated if it comes out from the optimization algorithm.
V. S
IMULATIONS RESULTS AND DISCUSSIONS
In this section, we evaluate the performances of the set
of studied estimators described in section III and IV through
Monte Carlo simulations. The different steps of the simulation
are the following:
1) K random ANs and one targeted MS are uniformly
drawninanareaofl ×lm
2
.
2) Different path losses (L L
0
) are computed for each
link k between the MS and the k
th
AN. For each link,
log normal shadowing model is applied with appropriate
n
p
, λ and σ
sh
. Table III shows the used parameters for
indoor and outdoor scenarios respectively.
3) The four different estimators are then evaluated for three
different scenarios:
Indoor.
Outdoor.
Indoor/Outdoor.
Parameters Indoor Outdoor
n
p
1.6to1.8 2to4.0
λ (m) 0.12 0.333
σ
sh
(dB) 2to5 2to5
l (m) 15 1000
TABLE III: List of radio parameters used in simulations for both
indoor and outdoor scenarios.
table
All simulations have been done with a number of trials
equal to N
T rial
= 300. For each studied scenario, the
correspondent figure (Fig. 2 to Fig. 4 respectively) compares
the cumulative density functions of four estimation schemes
with respect to the positioning error in order to suggest the
best estimation scheme.
The Fig. 2 and Fig. 3 are obtained respectively for indoor
and outdoor scenarios with the parameters described in Table
III. These figures show that the indirect estimation scheme
based on the mode estimator for r anges performs better than
those usually used based on median and mean estimators.
Moreover, these figures suggest that the new proposed direct
estimator performs better than direct schemes in the two
different cases (indoor and outdoor). Thus, we believe that
the direct RSS-based estimation scheme of MS’s position
may enhances the positioning accuracy.
In order to show the reliability of this new direct estimator
even in the case of hybrid RSS fusion, we carried simulations
in a typical 4G scenario where the MS can be connected
conjointly to cellular BSs and wireless APs. The Fig. 4 shows
the performances of different estimators for this scenario with
l = 1000 m. This figure is obtained by reproducing the same
simulations conditions assumed in Fig. 3 but with adding two
indoor links into a square of l =15m. The position of MS is
chosen randomly in the sqaure 15 × 15 m
2
. This is done by
respecting the different assumed parameters (n
p
, λ and σ
sh
)
for indoor scene given by Table III for each additional link.
This figure emphasizes the expected conclusions and shows
that the new proposed estimator enhances the performances
of hybrid RSS-based localization.
Comparison between Fig. 3 and Fig. 4 shows that the
enhancement performed by the direct estimator, after adding
indoor links, is major than the enhancement performed in the
case of indirect estimators. These first constatations suggest
that the proposed direct estimator is more reliable when
hybrid RANs are used. Furthermore in this type of scenarios,
estimators may experience short and long range links at the
same time. In this case, the precisions of estimated distances
from RSS observables can be very different as explained in
Fig. 1. We believe that the direct estimator is not influenced
by these imprecisions because it uses RSS observables directly
without going through ranges estimations.
VI. C
ONCLUSION
In this paper, we studied hybrid RSS-based localization
estimators assuming a path loss log normal shadowing model.
We distinguished direct from indirect schemes. Indirect esti-
mation schemes consist in two steps: estimation of ranges from
RSS using mean, median or mode estimators; and estimation
of location using weighted least square approximation on
previously estimated ranges. We showed that estimation of
ranges from RSS and consequently positioning accuracy can
be enhanced using mode estimator rather than median or mean
estimators usually used in past studies. Furthermore, a new
direct scheme for location estimation from RSS is proposed
and analyzed. This new estimator performs better than indirect
schemes and may enhances positioning accuracy using hybrid
RSS observables coming from different radio access networks.
Next step will be to evaluate performance in more realistic
scenarios and especially by using more realistic path loss
model with adequate parameters.

Citations
More filters

Proceedings Article
27 Apr 2011
TL;DR: This paper presents a simulation study of nonhybrid and hybrid localization techniques using RSSI, TOA, and TDOA location dependent parameters to show the importance of hybrid data fusion for localization.
Abstract: This paper presents a simulation study of nonhybrid and hybrid localization techniques using RSSI, TOA, and TDOA location dependent parameters. Maximum likelihood and weighted least squares are considered and developed for both non-hybrid and hybrid cases. Monte-Carlo simulations using realistic radio parameters extracted from an ultra wide band measurement campaign are carried out in order to assess the performances of different techniques and to show the importance of hybrid data fusion for localization.

50 citations


Cites background or methods from "Enhancing Positioning Accuracy thro..."

  • ...By deriving these likelihood functions, we obtain easily the different ML estimators for respectively RSSI, TOA, and TDOA [5], [6]....

    [...]

  • ...This paper considered non-hybrid and hybrid localization techniques using RSSI, TOA, and TDOA....

    [...]

  • ...Assuming Gaussian models independence between considered LDP measurements, the likelihood functions are given receptively for RSSI, TOA, and TDOA by [5], [6]: ⎧⎪⎪⎪⎪⎪⎨ ⎪⎪⎪⎪⎪⎩ fRSSI(X) = p∏ k=1 1√ 2πdkSk e − (ln dk−Mk) 2 2S2 k fTOA(X) = q∏ k=p+1 1√ 2πσk e − (cτk−dk) 2 2σ2 k fTDOA(X) = K∏ k=q+2 1√ 2πσk(q+1) e − (cτk(q+1)−dk(q+1))2 2σ2 k(q+1) (10) where Sk and Mk are defined for each k respectively by [5]: Sk = −σsh k ln 10 10np (11) Mk = (P0 − Pk) ln 10 10np + ln d0 (12) By deriving these likelihood functions, we obtain easily the different ML estimators for respectively RSSI, TOA, and TDOA [5], [6]....

    [...]

  • ...The use of WLS technique is quite different for RSSI, TOA, and TDOA....

    [...]

  • ...CDFs of positioning error using WLS and ML estimators applied on the fusion of RSSI, TOA, and TDOA. fusion of LDPs on positioning accuracy....

    [...]


Journal ArticleDOI
TL;DR: This survey summarizes and analyzes the existing fusion-based positioning systems and techniques from three characteristics, which consists of three fusion characteristics: source, algorithm, and weight spaces, and discusses their lessons, challenges, and countermeasures.
Abstract: Demands for indoor positioning based services (IPS) in commercial and military fields have spurred many positioning systems and techniques. Complex electromagnetic environments (CEEs) may, however, degenerate the accuracy and robustness of some existing single systems and techniques. To overcome this drawback, fusion-based positioning of multiple systems and/or techniques have been proposed to revamp the positioning performance in CEEs. In this paper, we survey the fusion-based indoor positioning techniques and systems from seminal works to elicit the state of the art within our proposed unified fusion-based positioning framework, which consists of three fusion characteristics: source, algorithm, and weight spaces. Different from other surveys, this survey summarizes and analyzes the existing fusion-based positioning systems and techniques from three characteristics. Meanwhile, discussions in terms of lessons, challenges, and countermeasures are also presented. This survey is invaluable for researchers to acquire a clear concept of indoor fusion-based positioning systems and techniques and also to gain insights from this survey to further develop other advanced fusion-based positioning systems and techniques in the future.

44 citations


Cites background from "Enhancing Positioning Accuracy thro..."

  • ...GSM, GPS, and Bluetooth [91], [103]–[105] because most wireless receivers can provide RSS measurements....

    [...]


Proceedings ArticleDOI
19 Mar 2009
TL;DR: This paper combines ToA and RSS and proposes a new estimator of ranges from RSS observables assuming a path loss model and a new ML estimator is developed to merge different ranges with different variances.
Abstract: In this paper, we exploit the concept of data fusion in UWB (Ultra Wide Band) localization systems by using different location-dependent observables We combine ToA (Time of Arrival) and RSS (Received Signal Strength) in order to get accurate positioning algorithmsWe assume that RSS observables are usually available and we study the effect of adding ToA observables on the positioning accuracy The proposed architecture of Hybrid Data Fusion (HDF) is based on two stages: Ranging using RSS and ToA; and Estimation of position by the fusion of estimated ranges In the first stage, we propose a new estimator of ranges from RSS observables assuming a path loss model In the second stage, a new ML estimator is developed to merge different ranges with different variances In order to evaluate these algorithms, simulations are carried out in a generic indoor environment and Cramer Rao Lower Bounds (CRLB) are investigated Those algorithms show enhanced positioning results at reasonable noise levels

31 citations


Patent
17 Oct 2011
Abstract: The present invention relates to a method for positioning a node within a wireless sensor network in which each node measures the RSSs from its neighboring nodes (310). Path loss parameters of a channel between a regular node and a neighboring node are first obtained (320). Distances separating each regular node from its neighboring nodes are then estimated (340) on the basis of the measured RSSs, the allocated path loss parameters. Each distance estimate is further corrected (340) by a systematic bias depending upon the actual sensitivity of the receiver. The positions of the regular nodes are estimated (350) from the distance estimates thus corrected.

21 citations


Journal ArticleDOI
TL;DR: A cost-effective hybrid analog digital system to estimate the Direction of Arrival (DoA) of WiFi signals using the so-called digital monopulse function, which can be applied to all WiFi standards and other Internet of Things narrowband radio protocols, such as Bluetooth Low Energy or Zigbee.
Abstract: We present a cost-effective hybrid analog digital system to estimate the Direction of Arrival (DoA) of WiFi signals. The processing in the analog domain is based on simple well-known RADAR amplitude monopulse antenna techniques. Then, using the received signal strength indicator (RSSI) delivered by a commercial MiMo WiFi cards, the DoA is estimated using the so-called digital monopulse function. Due to the hybrid analog digital architecture, the digital processing is extremely simple, so that DoA estimation is performed without using IQ data from specific hardware. The simplicity and robustness of the proposed hybrid analog digital MiMo architecture is demonstrated for the ISM 2.45 GHz WiFi band. Also, the limitations with respect to multipath effects are studied in detail. As a proof of concept, an array of two MiMo WiFi DoA monopulse readers is distributed to localize the two-dimensional position of WiFi devices. This cost-effective hybrid solution can be applied to all WiFi standards and other Internet of Things narrowband radio protocols, such as Bluetooth Low Energy or Zigbee.

16 citations


References
More filters

Book
01 Jan 2005

9,031 citations


MonographDOI
01 Jan 2005

5,313 citations


Journal ArticleDOI
01 Nov 2007
TL;DR: Comprehensive performance comparisons including accuracy, precision, complexity, scalability, robustness, and cost are presented.
Abstract: Wireless indoor positioning systems have become very popular in recent years. These systems have been successfully used in many applications such as asset tracking and inventory management. This paper provides an overview of the existing wireless indoor positioning solutions and attempts to classify different techniques and systems. Three typical location estimation schemes of triangulation, scene analysis, and proximity are analyzed. We also discuss location fingerprinting in detail since it is used in most current system or solutions. We then examine a set of properties by which location systems are evaluated, and apply this evaluation method to survey a number of existing systems. Comprehensive performance comparisons including accuracy, precision, complexity, scalability, robustness, and cost are presented.

3,865 citations


"Enhancing Positioning Accuracy thro..." refers methods in this paper

  • ...Fingerprinting with RSS refers to the type of algorithms that first collect RSS fingerprints of a scene and then estimate the location of the MS by matching on-line measurements with the closest location fingerprints [5]....

    [...]


Proceedings ArticleDOI
05 Dec 2005
Abstract: Empirical propagation models have found favour in both research and industrial communities owing to their speed of execution and their limited reliance on detailed knowledge of the terrain. Although the study of empirical propagation models for mobile channels has been exhaustive, their applicability for FWA systems is yet to be properly validated. Among the contenders, the ECC-33 model, the Stanford University Interim (SUI) models, and the COST-231 Hata model show the most promise. In this paper, a comprehensive set of propagation measurements taken at 3.5 GHz in Cambridge, UK is used to validate the applicability of the three models mentioned previously for rural, suburban and urban environments. The results show that in general the SUI and the COST-231 Hata model over-predict the path loss in all environments. The ECC-33 models shows the best results, especially in urban environments.

542 citations


"Enhancing Positioning Accuracy thro..." refers background in this paper

  • ...The total pathloss at a distance, d, will then be L, often modelled as [10]:...

    [...]


Journal ArticleDOI
TL;DR: This study demonstrates that RSS can be used to localize a multi-hop sensor network, and it is shown that this result is highly sensitive to subtle environmental factors such as the grass height, radio enclosure, and elevation of the nodes from the ground.
Abstract: Radio signal strength (RSS) is notorious for being a noisy signal that is difficult to use for ranging-based localization. In this study, we demonstrate that RSS can be used to localize a multi-hop sensor network, and we quantify the effects of various environmental factors on the resulting localization error. We achieve 4.1m error in a 49 node network deployed in a half-football field sized area, demonstrating that RSS localization can be a feasible alternative to solutions like GPS given the right conditions. However, we also show that this result is highly sensitive to subtle environmental factors such as the grass height, radio enclosure, and elevation of the nodes from the ground.

352 citations


"Enhancing Positioning Accuracy thro..." refers background in this paper

  • ...1, changes in RSS due to distance become small relative to noise, even if the level of noise remains the same over distance [11]....

    [...]


Frequently Asked Questions (1)
Q1. What are the contributions mentioned in the paper "Enhancing positioning accuracy through direct position estimators based on hybrid rss data fusion" ?

In this paper, localization based on Received Signal Strength ( RSS ) is investigated assuming a path loss log normal shadowing model. The authors show that the performances of this type of schemes depend on the used estimator in the first step. The authors suggest that typical median estimator must be replaced by maximum likelihood estimator ( mode ) to enhance the positioning accuracy.