sensors

Article
An RFID Indoor Positioning Algorithm Based on
Bayesian Probability and K-Nearest Neighbor
He Xu 1,2, * ID
, Ye Ding 1,2 , Peng Li 1,2 , Ruchuan Wang 1,2 and Yizhu Li 3
1 School of Computer Science, Nanjing University of Posts and Telecommunications, Nanjing 210023, China;
[email protected] (Y.D.); [email protected] (P.L.); [email protected] (R.W.)
2 Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks, Nanjing 210003, China
3 Department of Electrical and Computer Engineering, New Jersey Institute of Technology,
Newark, NJ 07102, USA; [email protected]
* Correspondence: [email protected]; Tel.: +86-25-8586-6354

Received: 15 June 2017; Accepted: 1 August 2017; Published: 5 August 2017

Abstract: The Global Positioning System (GPS) is widely used in outdoor environmental positioning.
However, GPS cannot support indoor positioning because there is no signal for positioning in an
indoor environment. Nowadays, there are many situations which require indoor positioning, such
as searching for a book in a library, looking for luggage in an airport, emergence navigation for
fire alarms, robot location, etc. Many technologies, such as ultrasonic, sensors, Bluetooth, WiFi,
magnetic field, Radio Frequency Identification (RFID), etc., are used to perform indoor positioning.
Compared with other technologies, RFID used in indoor positioning is more cost and energy efficient.
The Traditional RFID indoor positioning algorithm LANDMARC utilizes a Received Signal Strength
(RSS) indicator to track objects. However, the RSS value is easily affected by environmental noise and
other interference. In this paper, our purpose is to reduce the location fluctuation and error caused by
multipath and environmental interference in LANDMARC. We propose a novel indoor positioning
algorithm based on Bayesian probability and K-Nearest Neighbor (BKNN). The experimental results
show that the Gaussian filter can filter some abnormal RSS values. The proposed BKNN algorithm
has the smallest location error compared with the Gaussian-based algorithm, LANDMARC and an
improved KNN algorithm. The average error in location estimation is about 15 cm using our method.

Keywords: indoor positioning; RFID; Bayesian probability; K-Nearest Neighbor

1. Introduction
Location-based services (LBS) are provided by a number of commercial companies, and it
improves user satisfaction and offers convenience. Many LBS applications need to know the detailed
location of objects. Over the years, many systems have supported automatic positioning, such
as the Global Positioning System (GPS), widely used in outdoor environments all over the world.
However, GPS cannot support indoor positioning because there is no signal inside buildings. In order
to meet the requirements for inside buildings, many indoor positioning technologies are proposed
by researchers. Ultrasonic [1], sensor [2], Bluetooth [3], WiFi [4] and magnetic field [5] technologies
are used to perform indoor positioning.
Yayan Ugur et al. proposed an ultrasonic based indoor positioning system [6], which uses only
ultrasonic signals and calculates the position of the mobile platform with centimeter-level accuracy.
The transmitters are at fixed known positions and the time difference of arrival (TDOA) of the receiver,
which is embedded on the mobile unit, is calculated to get the mobile unit’s position. The system
is cost effective because it uses the existing Local Area Network of the building.
Deng Zhi-An et al. proposed a novel position transition detection algorithm [7], which allows
the user heading estimation to be pervasive and suitable for daily use and for indoor pedestrian

Sensors 2017, 17, 1806; doi:10.3390/s17081806 www.mdpi.com/journal/sensors

Sensors 2017, 17, 1806 2 of 17

navigation using the built-in inertial sensors on a smartphone. It can be used to avoid the confusion
between position transition and user turn during pedestrian walking. The results show that the
proposed approach can automatically detect carrying positions with high accuracy.
Li Honggui proves that the nonlinear least square method is suitable for parameter estimation
of Bluetooth signal propagation and provides the theoretical details for 3D indoor position with
a Bluetooth device [8]. The results of simulation and hardware experiments show that the average of
absolute precision of location estimation is at meter level.
Han-Sol Kim et al. proposed an indoor positioning system [9] with a particle filter system where
the weights of particles are updated by multiple magnetic sensors and three magnetic field maps:
a horizontal intensity map, a vertical intensity map and a direction information map. The system
integrates magnetic field map navigation and an encoder system. The results show that the estimated
position of the proposed system is more accurate than that of previous systems.
Nascimento Hitalo J.B. et al. proposed a positioning algorithm [10] based on Bayes inference
to locate objects in 3D WLAN indoor environments. This is a fingerprint technique and the average
positioning error is about three meters.
With the increasing number of mobile robots controlled by a mobile device, Chao Chun-Tang
et al. proposed a visual control interface for a mobile robot with a single camera to estimate the 3D
position of a target [11]. WiFi is used to transmit the control signal and video from the mobile robot.
The presented 3D estimation method is based on triangulation.
Radio frequency identification (RFID) [12] is the key technology necessary to realize the Internet
of Things (IoT) and cyber-physical systems (CPS), widely used in health monitors [13,14], smart
homes [15], smart cities [16], vehicle location [17], construction [18], supply chain management [19]
and object tracking [20], etc. RFID has the advantages of low cost, long life, low power consumption
and easy deployment, which attracts many researchers to use it in indoor environments [21–23].
Compared with ultrasonic, WiFi and Bluetooth, RFID used in indoor positioning is more cost and
energy efficient. The typical RFID systems consists of tags, readers and a back-end computer system.
There are two types of tags: active and passive, where the active tag has a battery and can send the
information at hundreds of meters, and the passive tag depends on the emitted energy of the reader’s
antenna instead of using a battery to transmit information. There are four different RFID frequency
bands [24]: low frequency (LF)—125~134 kHz, high frequency (HF)—13.56 MHz, ultra-high frequency
(UHF)—433~956 MHz and microwave frequency (MF)—2.45 GHz. Ni Lionel M. et al. first used active
RFID with the concept of reference tags for indoor location sensing [25], called LANDMARC, and as
they used K-nearest reference tags’ coordinates to locate unknown tags, it is also called a K-Nearest
Neighbor (KNN) algorithm [26]. Benelli Giuliano et al. uses LF RFID to track different typologies of
pebbles on beaches [27]. Mi Jian et al. designed an HF-band RFID system with multiple readers and
passive tags for indoor mobile robot self-localization [28]. Own Chung-Ming et al. uses UHF RFID
technologies for real-time bus recognition in the Taipei bus station [29], where the active RFID tag on
the bus is combined with the infrared receiver. Hsu Chien-Chang et al. proposed a sensor-assisted
RFID-based indoor tracking system [23] which uses active RFID technology to identify the location of
elderly people living alone. The active RFID system may exhaust much more energy than the passive
RFID system. The LF RFID system has a short reading range at centimeter-level and is not suitable for
large-scale location usage.
There are two types of Received Signal Strength indicator (RSS, many references also refer
to it as RSSI) based indoor positioning algorithms in RFID systems: triangulation and reference
tags. The triangulation method uses the geometry of circles, spheres or triangles to determine the
locations of the target by estimating distances. The following methods are usually used to estimate
the distance in triangulation based position systems: Time of Arrival (TOA) [30] and Angle of Arrival
(AOA) [31]. The authors [32] present a novel passive RFID localization algorithm based on elliptical
trilateration in smart home environments, where they obtained an average error of 16.08 cm for all
objects. The reference tags method first requires the building of a database of known RSS signals of
Sensors 2017, 17, 1806 3 of 17

the RFID tags’ locations. It then uses a positioning algorithm to calculate the target tag’s location.
In the literature [33], the authors present a maximum intersected tags method to estimate the target’s
location. The results show that the location error is about 30~60 cm. Table 1 gives the comparisons of
different positioning methods. The COTS reader is a commercial off-the-shelf reader.

Table 1. Comparisons of different positioning methods.

Index Time Energy COTS

Accuracy Cost
Positioning Methods Consumption Consumption Reader
Fingerprint [10,33] High Low High Low Yes
Reference tags [25] Normal High Normal Normal Yes
AOA [31] High Normal Normal High Yes
TDOA [6] Low Low Normal Low No

Due to the low cost and easy deployment of passive UHF RFID it is a widely used and
researched positioning method. Buffi, A. et al. proposed a phase-based technique for localization and
tracking of UHF RFID tags moving along a conveyor belt [34], and the numerical results show that
a centimeter-level accuracy in the tag position estimation can be achieved even in a rich multipath
environment. In order to overcome the limitations of current RFID systems, Guidi, F. et al. proposed
the joint adoption of RFID and ultra-wideband (UWB) tag technologies [35,36], which are introduced
to reduce the positioning error thanks to the large bandwidth involved. Arnitz D. et al. give a proof of
concept for UWB-based ranging in passive UHF RFID [37], which is robust to multipath propagation.
UWB RFID will offer high accuracy localization capabilities in next generation RFID systems. However,
the UWB RFID is limited in the range of reader-tag communication [38] and it is costly, as the high
accuracy localization requires three readers. To overcome such limitations, Guidi, F. et al. investigated
the possibility of jointly adopting millimeter-wave (mmW) technology with passive RFID [39], and
the results show that the proposed scheme can outperform the UWB one as it requires a lower number
of pulses to achieve the same reader-tag distance and consequently a reduced complexity.
LANDMARC [25] is the first method to use reference tags to locate the target, and its positioning
accuracy is about 72 cm. The KNN algorithm [25] is a kind of references tag positioning method,
mainly used to estimate the location according to the measured RSS compared with the known RSS
value recorded in the database. The KNN is widely used and improved in indoor environment
positioning. Li D. in literature [40] proposed a feature-scaling-based k-nearest neighbor (FS-kNN)
algorithm for achieving improved localization accuracy, while FS-kNN can achieve an average location
error as low as 1.70 m in WiFi fingerprint-based indoor positioning systems. An improved KNN
algorithm was introduced by Liang X. et al. which is beneficial to location estimation in a real GSM
network [41]. In order to meet the demand of localization precision, LANDMARC is improved as well.
In order to find the k value which leads to the best accuracy in each environment, an adaptive KNN
algorithm is proposed for RFID location systems [42]. Because the multipath effect and the various
indoor environments significantly affect the positioning accuracy, the overall accuracy of LANDMARC
fluctuates. In order to solve this problem, Liu X. et al. proposed LANDMARC with an improved KNN
algorithm for RFID location systems whose location accuracy is increased by 10.1% on average [43].
This paper mainly focuses on the passive UHF RFID-based indoor positioning method. Inspired
by the WiFi indoor positioning algorithm [10] and the KNN algorithm [26], and inspired by Sun Y.
and Gu F. who use the sparse signal to descript all the information of the original signal with high
probability in the literature [44], we propose an RFID indoor positioning algorithm based on Bayesian
probability and K-Nearest Neighbor, using a Gaussian filter to filter the abnormal Received Signal
Strength (RSS) value and using the proper k value and the Bayesian estimation method to improve
the accuracy of the location. Rather than WiFi technology [10] and active RFID technology [25], we use
passive UHF RFID technology to perform indoor positioning resulting in a lower location error of about
15 cm. When comparing the proposed algorithm with the traditional LANDMARC algorithm and
other related algorithms, the results show that our algorithm performs with a higher location accuracy.
 Sensors 2017, 17, 1806 4 of 17

algorithm
Sensors and
2017, 17, other
1806 related algorithms, the results show that our algorithm performs with a higher 4 of 17
location accuracy.
The rest of the paper is organized as follows: In Section 2, we examine the RFID signal
The rest of
characteristic of the
RSSpaper is organized
and phase. Section as follows:
3 gives In Sectiontheorem
the Bayesian 2, we examine the algorithm.
used in our RFID signal In
characteristic of RSS and phase. Section 3 gives the Bayesian theorem used
Section 4, an RFID indoor positioning algorithm is proposed where a Gaussian filter is used in our algorithm.
to filter
In Section 4,
abnormal an values
RSS RFID indoor positioning
and Bayesian algorithm
estimation is proposed
is used where
to improve a Gaussian
positioning filter isSection
accuracy. used to5
filter abnormal RSS values and Bayesian estimation is used
presents the experiments. Finally, Section 6 concludes the paper. to improve positioning accuracy. Section 5
presents the experiments. Finally, Section 6 concludes the paper.
2. RFID Signal Characteristic
2. RFID Signal Characteristic
In the following, we use the commodity off-the-shelf RFID reader—Impinj R420 RFID reader
In the following, we use the commodity off-the-shelf RFID reader—Impinj R420 RFID reader and
and EPC Gen2 UHF RFID tags to perform the test for the signal characteristic of RSS and phase.
EPC Gen2 UHF RFID tags to perform the test for the signal characteristic of RSS and phase.
2.1. RSS
2.1. RSS
RSS is an important eigenvalue of RFID signals, which can be obtained by the RFID reader
RSS is an important eigenvalue of RFID signals, which can be obtained by the RFID reader
reading the RFID tags. Through testing, we found that the RSS value has a linear relationship with
reading the RFID tags. Through testing, we found that the RSS value has a linear relationship with
the distance between the antenna of the RFID reader and the RFID tags. In Figure 1, we do the test
the distance between the antenna of the RFID reader and the RFID tags. In Figure 1, we do the test with
with a continuously increasing distance from 0 to 120 cm by slowly moving an electric toy car
a continuously increasing distance from 0 to 120 cm by slowly moving an electric toy car equipped
equipped with an RFID tag further away, step by step. At the same time, the RFID reader reads the
with an RFID tag further away, step by step. At the same time, the RFID reader reads the tag and
tag and records the RSS value. When the tag is far enough away from the reader’s antenna, the RSS
records the RSS value. When the tag is far enough away from the reader’s antenna, the RSS value does
value does not change significantly. From Figure 1, we can see that when the distance is at 120 cm,
not change significantly. From Figure 1, we can see that when the distance is at 120 cm, the RSS value
the RSS value is stable.
is stable.

Figure 1. The RSS value changes according to distance.

Figure 1. The RSS value changes according to distance.

In
In addition,
addition, we
we put
put the tag approximately
the tag approximately 0.6 0.6 mm in
infront
frontofofthe
thereader’s
reader’santenna,
antenna,asasshown
shown in
in Figure 2. At the same time, we recorded the RSS value. When there are shelters
Figure 2. At the same time, we recorded the RSS value. When there are shelters to pass through to pass through
between
between the reader and
the reader andthethetag,
tag,thethe
RSSRSS value
value fluctuates
fluctuates greatly,
greatly, whichwhich is shown
is shown in Figure
in Figure 3. Num- 3.
Num-Announces are the cumulative sums of numbers of a tag read successfully
Announces are the cumulative sums of numbers of a tag read successfully by the RFID reader, whichby the RFID reader,
which is normally
is normally at aof
at a rate rate of approximately
approximately 40 readings
40 readings per minute.
per minute. From From Figure
Figure 3, we
3, we cancan
see see
thatthat
the
the fluctuations are generated when the Num-Announces are about 100, 320, 500 and
fluctuations are generated when the Num-Announces are about 100, 320, 500 and 715, respectively. 715, respectively.
This
This isisbecause
because wewe repeatedly
repeatedly putput aahand
handin inthe
thespace
spacebetween
between the thereader
readerand
andthethetag,
tag,causing
causing
interference
interferencewith
withthe
thewireless
wirelesscommunication
communicationwhich whichinfluences
influencesthe themeasured
measuredRSS.
RSS.
Sensors 2017, 17, 1806 5 of 17
Sensors 2017,
Sensors 17,17,
2017, 1806
1806 5 of 17 17
5 of

Figure
Figure 2.
Figure Obtaining
2. 2. the
Obtaining
Obtaining RSS
the
the value.
RSS
RSS value.
value.

Figure 3. The RSS value in an interference environment.

Figure 3. 3.
Figure The RSS
The value
RSS in in
value anan
interference environment.
interference environment.

2.2.
2.2. Phase
Phase
2.2. Phase
The
TheThephase
phase
phase is is
is another
another
another important
important
important eigenvalue
eigenvalue
eigenvalue ofof
of RFID
RFID
RFID signals
signals
signals which
which
which the
the RFID
RFID
the RFID reader
reader
reader antenna
antenna
antenna can
cancan
read
read
read from
from
from the
thethetag
tag tag [45].
[45].
[45]. Similarly,
Similarly,
Similarly, we
wewe place
place
place the
thethetagtag
tag at
atat 0.6
0.60.6 mmm as
asas shown
shown
shown ininin Figure
Figure
Figure2. 2.2.
AtAt At
the the
same
the same
same time, time,
time, wewe
werecord
record
record the
thethe
phasephase
phase value.
value.
value. AsAs As Figure
Figure
Figure 4 shows,
4 shows,
4 shows, if if
the if distance
the the distance
distance between
between
between thethe the
tag tagtag
and and
andthethe the
readerreader
reader is is is fixed,
fixed,
fixed, thethe
the phase
phase
phase value
value
value changes
changes
changes from from
from 1.50 1.50
1.50 to1.56
toto1.56 1.56and and
and the the
the mean
mean
mean of
ofof the
the phase
phaseisisis1.53,
thephase 1.53, which
whichisisiscaused
1.53,which caused
causedby bybythe
thethe
multipath
multipath
multipath effect
effect
effect and
andand the
thethe various
various
various noises
noises
noises in
inin the
thethe indoor
indoor
indoor environment.
environment.
environment. However,
However,
However, wewewecan can
can consider
consider
consider 1.531.53
1.53asas
asthe
thethe phase
phase
phase value
value
value ininin
thethe
the experiment
experiment
experimentthrough through
throughsimple simple
simpledata data processing.
dataprocessing.
processing.We We hold
Wehold
holdthe the view
theview that
viewthat this
thatthisthis
characteristic
characteristic
characteristic ofof
of phase
phase
phase inin
in aa fixed
fixed
a fixed distance
distance
distance belongs
belongs
belongs to
totoaa stable
stable
a stablestatus.
status.
status.
We
We Weput
putputthe
the
the RFID
RFID
RFID tag
tag taginininfront
front
front ofof
of the
thethe antenna
antenna
antenna ofofthe
of the
the RFID
RFID
RFID reader
reader
reader and and
and keptkept
away
kept away
away from from
from it it itwhile
while while
wewe
we
usedused
used the
thethe
RFID RFID
RFID readerreader
reader toto to
readreadread
thethethe
phase phase
phase value value
value atat theat
the the
same same
same time.time.
time.AsAs As
shown shown
shown inin in
Figure Figure
Figure 5, 5,
wewe5, we
found
foundfound
that
that
that
the the
thephase phase
phasevalue value
valuehas has a
hasa alinearlinear relationship
linearrelationship
relationshipwith with
withthe the distance
thedistance
distanceand and is cyclical.
andis iscyclical. In addition,
cyclical.InInaddition,
addition,the the phase
thephase
phase
difference
difference
difference [46]
[46]
[46]ofof
of tag
tagtagsensitivity
sensitivity
sensitivity makes
makes
makes it
it hard
hard
it hard to
toto use
useusepassive
passive
passiveRFID
RFID
RFID for large-scale
forfor
large-scale
large-scale location
location
location usage.
usage.
usage.
 Sensors 2017, 17, 1806 6 of 17
Sensors 2017, 17, 1806 6 of 17
Sensors 2017, 17, 1806 6 of 17

Figure4.4.The
Figure Thephase
phasevalue
valueisisstable
stableininthe
thefixed
fixeddistance.
distance.
Figure 4. The phase value is stable in the fixed distance.

Figure 5. The phase value is cyclical and linear with changeable distance.
Figure 5. The phase value is cyclical and linear with changeable distance.
Figure 5. The phase value is cyclical and linear with changeable distance.
Thus,
Thus,ififwe
weuseusethe
theRFID
RFIDsignal
signal to
to carry
carry out indoor positioning,
out indoor positioning, wewemust
mustovercome
overcomethethenoise
noisein
inindoor
indoor environment
environment
Thus, if we useandand the
the the
RFIDcycle
cycle characteristic
characteristic
signal of the
the phase.
of indoor
to carry out phase. In
In this
thispaper,
positioning,paper, wewemainly
we must mainlysolve
solve
overcome the
the noise
thenoise
noisein
problem
problem which influences
which influences
indoor environment the location
thecycle
and the accuracy.
location In the following,
accuracy.ofInthe
characteristic the we
following,
phase. will
In thiswe use a
willwe
paper, Gaussian
usemainly filter
a Gaussian to filter
filter
solve the to
noise
abnormal
filter
problem RSS
abnormal
whichvalues
RSS and usethe
values
influences Bayesian
use estimation
andlocation Bayesian with thefollowing,
estimation
accuracy. In the KNN
withalgorithm
thewe KNN toalgorithm
will improve
use positioning
to improve
a Gaussian filter to
accuracy in RFID
positioning
filter abnormal systems.
accuracyRSSinvalues
RFID systems.
and use Bayesian estimation with the KNN algorithm to improve
positioning accuracy in RFID systems.
3.3.Bayes
BayesTheorem
Theorem
3. Bayes Theorem conditional probability in statistics is:
The formula
The formulafor for conditional probability in statistics is:
The formula for conditional probability in statistics is:
P( ABP(AB)
) = P(=AP(A)P(B|A)
) P( B| A) = P=( B
P(B)P(A|B),
) P ( A | B ), (1)
(1)
P(AB) = P(A)P(B|A) = P(B)P(A|B), (1)
That is, the probability 𝑃(𝐴𝐵) that event A and event B occur at the same time is equal to the
That is, the probability P( AB) that event A and event B occur at the same time is equal to
That is,𝑃(𝐴)
probability
the probability the ) that event𝑃(𝐴𝐵)
that event A
P( Aprobability
occurs multiplied
that
A occurs event A and
multiplied
by the
by event B occur atP𝑃(𝐵|𝐴)
probability
the probability the
( B| same
that
A) that
event
time
event
B occurs
is equal to the
B occurs
under
probability 𝑃(𝐴)
the probabilitythat of the A
event occurrence
occurs of event by
multiplied A. the
Or probability
the probability 𝑃(𝐵|𝐴)𝑃(𝐴𝐵)thatisevent
equalB to the
occurs
under the probability of the occurrence of event A. Or the probability P( AB) is equal to the probability
under the 𝑃(𝐵)
probability that event
probability B occurs
of the multiplied
occurrence of eventby A.
theOrprobability
the probability𝑃(𝐴|𝐵)𝑃(𝐴𝐵)
that event A occurs
is equal to the
under
probability 𝑃(𝐵) that
the probability of event
the occurrence
B occurs ofmultiplied
event B. From Equation
by the probability(1), we𝑃(𝐴|𝐵)
can get:that event A occurs
under the probability of the occurrence of event B. From Equation (1), we can get:
Sensors 2017, 17, 1806 7 of 17

P( B) that event B occurs multiplied by the probability P(A|B) that event A occurs under the probability
of the occurrence of event B. From Equation (1), we can get:

P( B) P( A| B)
P( B| A) = (2)
P( A)

Assume that event B is a probability space which is composed of independent events, then P( A)
can be expanded by the full probability formula:

P( A) = P( A| B1 ) P( B1 ) + P( A| B2 ) P( B2 ) + · · · + P( A| Bn ) P( Bn ) (3)

Thus, the Bayesian formula can be expressed as:

P( A| Bi ) P( Bi )
P( Bi | A) = (4)
P( A| B1 ) P( B1 ) + P( A| B2 ) P( B2 ) + · · · + P( A| Bn ) P( Bn )

P( Bi | A) refers to the posterior probability, but P( A| Bn ) P( Bn ) refers to the prior probability, and
P( Bi ) refers to the basis probability. The Bayesian formula also can be expressed as:

P( A| Bi ) P( Bi )
P( Bi | A) = (5)
∑in=1 P( Bi ) P( A| Bi )

Normally, P( A| B) is not equal to P( B| A) . However, the two have a definite relationship,

where Bayesian law is the statement of this relationship. Bayesian law is the relationship of the
conditional probability and the marginal probability of random events A and B, which is shown in the
following formula:
P( B| A) P( A)
P( A| B) = ∝ L( A| B) P( A) (6)
P( B)
Where P( A| B) is the possibility of event A occuring in the case of event B having occurred, P( A)
is the prior probability (also called the marginal probability) of event A which is unaffected by any
factor of event B, P( A| B) is the conditional probability of event B having already occurred and is
also called the posterior probability of event A, P( B| A) is the conditional probability of event A
having already occurred and is also called the posterior probability of event B, and P( B) is the prior
P( B| A)
probability of event B which is unaffected by any factor of event A. In addition, sometimes P( B) is also
called the standardized likelihood. Then the Bayesian law can be expressed as: Posterior probability
= Standardized likelihood × Prior probability.
In summary, the parameters are being estimated as a random variable which is a specific
distribution form, and the probability density distribution f ( x |θ ) is transferred as the posterior
probability h( x |θ ). It first obtains the probability density distribution and then obtains the parameter
estimation value through mathematical expectations. The specific steps of the Bayesian estimate are
used in our method and listed as follows:

(1) Confirm the prior distribution π (θ ) of parameter θ.

(2) Obtain the joint probability density distribution f ( x |θ ) through joint probability density
distribution x = ( x1 , x2 , x3 , · · · , xn ).
(3) Obtain the posterior probability h( x |θ ) of parameter θ.
(4) Obtain the estimated value θ̂

4. Proposed Positioning Algorithm

The indoor environment is changeable because of the intensity of the population, the random
walking behavior of person, the path’s effect and so on. This will make the RSS value of the RFID
tag fluctuate around a certain value. The RSS value in the same location can be approximated
Sensors 2017, 17, 1806 8 of 17

by the Gaussian probability distribution, and the probability density function can be expressed as
the following:
1 ( x − µ )2
−
P( x ) = √ e 2σ2 (7)
σ 2π
where x represents the actual RSS value, µ = m1 ∑im=1 RSSIi represents the mean of the RSS value,
q
1 m 2
σ = m− 1 ∑i =1 ( RSSIi − µ ) and σ represents the standard deviation.
In order to overcome the noise in the indoor environment, we used a Gaussian filter to filter
the abnormal RSS value and used Bayesian estimation to improve positioning accuracy. The proposed
positioning algorithm based on Bayesian probability and KNN (BKNN) has the following three
processes: (1) KNN to select k reference tags, (2) the Gaussian filter process and (3) the Bayesian
estimation process. The details of algorithm BKNN are as follows:

• KNN

Suppose there are n readers and m reference tags in an RFID systems, and u target tags are need
to be positioned. The RSS vector of the target tags is s = (s1 , s2 , s3 , · · · , sn ), which indicates the RSS
value of the target tags detected by the reader i, i ∈ (1, n). For the reference tags, the corresponding
√
RSS vector is θ = (θ1 , θ2 , θ3 , · · · , θn ). The Euclidean distance is defined as Ej = θi − si , j ∈ (1, m),
which is based on the target tag and reference tag. The Euclidean distance E is used to measure the
proximity of the target tag and the reference tag, that is, while the reference tag is close to the target tag
the E value should be smaller. Therefore, when there are m reference tags, the E value of the target
tag p and all reference tags should be smaller. Accordingly, the E value vector of the target tag p
and all reference tags can be expressed as E = ( E1 , E2 , E3 , · · · , Em ), and sorted in order from small
to large which is recorded as E0 = ( E1 0 , E2 0 , E3 0 , · · · , Em 0 ). From this sequence, the k reference tags
are selected and the weighted average value is calculated according to the known tag position, and
the approximate position of the target tag can be estimated.

• Gaussian Filter Process

In order to make the collected RSS value close to the real value and thus improve the positioning
accuracy, we need to filter the fluctuating RSS value which is generated by the probability event
or noise interference. Take the RSS collection process as a normal distribution model. When we
calculate the RSS, assuming that it is a Gaussian distribution in which µ represents the mean of the RSS
value and σ represents the standard deviation. The result of the density function of x is expressed as
the following:
1 ( x − µ )2
−
f ( x ) = √ e 2σ2 (8)
σ 2π
q
2
where µ = m1 ∑im=1 RSSIi , σ = 1 m
m−1 ∑i =1 ( RSSIi − µ ) , RSSIi is the i-th RSS value, and m is the
measurement time. The greater the value of σ is, the better the smoothing effect of the Gaussian filter
is. We select the a probability is greater than 0.6 which is in the high probability generation area,
that is 0.6 ≤ f ( x ) ≤ 1, 0.15σ + µ ≤ x ≤ 3.09σ + µ. This range of RSS values is selected, and then
the geometric average is calculated which is regarded as the target tag’s RSS value.

• Bayesian Estimation Process

When calculating the target tag’s coordinate using the LANDMARC algorithm, a different k
value will produce a different coordinate. The measurement can be repeated k times. Suppose
the measurement is independent, according to Bayesian theory, the prior probability model and
the posteriori probability model can be set up. The prior probability model is used to estimate
the probability at k time of the target’s probable position at k − 1 time. After that, the posteriori
Sensors 2017, 17, 1806 9 of 17

probability model is set up according to the measurement value of the target’s position at k time, and
then the final position can be calculated. The details are shown in the following.
Let Zk = { Zi , i = 1, 2, 3, . . . , k} represents the sets of all measured k tags’ coordinates, and Zi
is the i-th tag’s measured coordinates. p( Zk | Xk ) is the probability of the unknown tag at Xk position
under the known set Zk , and p( Zk | Xk ) is also called as posteriori probability; p( Xk | Zk−1 ) is the prior
probability which represents the position estimation of unknown tag under the unknown set Zk .
Assuming that the measured values are independent of each other, we can get the following formula
according to Bayesian formula:

p( Xk | Zk ) ∝ p( Zk | Xk ) × p( Xk | Zk−1 ) (9)

The proposed Bayesian estimation based positioning algorithm is as follows:

(1) Determine the prior probability: p( Xk | Zk−1 ). Suppose Xk = ( x, y), Xk−1 = ( xk−1 , yk−1 ), we can
obtain the following formula:

D2
− 1
1 2σ2
p( Xk | Zk−1 ) = √ e 1 (10)
σ1 2π

where D1 is the distance between Xk and Xk−1 .

(2) Determine the posteriori probability: p( Zk | Xk ), which is shown in the following formula.

( D2 − Q)2
1 −
2σ22
p( Zk | Xk ) = √ e (11)
σ2 2π

where p( Zk | Xk ) is the probability of the unknown tag at Xk position and the measured
distance is D2 , Q represents the distance between the unknown tag and the reader’s antenna,
and σ2 represents the uncertainty variance of the measured distance. In our algorithm,
four reference tags around the target tags’ position are selected, where
q these positions are
A( a1 , b1 ), B( a2 , b2 ), C ( a3 , b3 ), D ( a4 , b4 ), then we can get the value of D2 = ( x − a i ) 2 + ( y − bi ) 2 .
(3) Calculate the probability distribution function of the unknown tag’s position. According to
Formulas (9)–(11), we can calculate the p( Xk | Zk ) which represents the probability distribution
function of the target tag’s location. We then get the following formula:

4  4
1
p( x, y) = ∏ pi (Xk |Zk ) = C σ1 σ2 2π
√ e f ( x,y) (12)
i =1

where
( q 2 )
4 4 h i
2 2 2 2
f ( x, y) = ∑ f i ( x, y) = − 21 ∑ 1
σ12
( x − x k −1 ) + ( y − y k −1 ) + 1
σ22
( x − a i ) + ( y − bi ) − q i (13)
i =1 i =1

In Equation (13), qi is the distance between the reference tag and the target tag. The value
of ( x, y) which could make p( x, y) to be the maximum value can be regarded as the target
position. The exponential function is monotonic, and we can find the ( x, y) that can make f ( x, y)
q the maximum. The target function f ( x, y) contains nonlinear factors, we define gi ( x, y) =
has
( x − ai )2 + (y − bi )2 and expand the approximate value (x 0 , y0 ) with Taylor expansion type, and
then we can get the following equation:
Sensors 2017, 17, 1806 10 of 17

x 0 − bi

∂gi ( x,y)
 mi = ∂x |( x,y) = ( x 0 , y0 ) = q
( x 0 − ai )2 +(y0 −bi )2




y 0 − bi

∂gi ( x,y)
∂y |( x,y) = ( x 0 , y0 ) =

 ni = q
( x 0 − ai )2 +(y0 −bi )2 (14)
 q
g i = g i ( x 0 , y 0 ) = ( x 0 − a i ) 2 + ( y 0 − bi ) 2





gi ( x, y) ≈ gi + mi ( x − x 0 ) + ni (y − y0 ) = mi x + ni y + gi − mi x 0 − ni y0



1 1
Let ci = gi − mi x0 − ni y0 − qi , w1 = σ21
, w2 = σ22
, we can get the following formula:

4 4
1 n h i o
f ( x, y) = ∑ fi (x, y) = − 2 ∑ w1 ( x − x k −1 ) 2 + ( y − y k −1 ) 2 + w2 ( m i x + n i y + c i ) 2 (15)
i =1 i =1

(4) Calculate the target tag’s position. What makes p( x, y) to be the maximum value can be
considered asSensors
the target tag’s location. Let the first derivative of Equation (15) equals
2017, 17, 1806 10 ofzero,
17 we can get
the following equation:
ì 4 4 4
(4w1 + w2 å mi 2 ) x + w2 (å ni mi ) y + w2 å ci mi - 4w1 xk -1 = 0
4ïï 4 i=1 4

i =1 i =1

 (4w1 + w2 ∑íï mi 24 ) x + w2 ( ∑ n4 i m2 i )y +4w2 ∑ ci mi − 4w1 xk−1 = 0 (16)


i =ï1w2 (å ni mi ) x + (4w1 i+=w12 å ni ) y + w2 å ci nii =

- 41w1 yk -1 = 0
4 î i =1
4 i =1
4 i =1 (16)
2( ∑
Thenwthe ni mofi ) x7,+
c (which 2∑
4w1 +is wthe ni 2 )yof+the ∑ ci n
w2target i− k −1 = 0 using
4wbe1 ycalculated

value position tag can


Equation (16).i=1 i =1 i =1

Then the 5.value

Experiments
of ( x, and Results is the position of the target tag can be calculated using Equation (16).
y) which
5.1. Experimental Settings
5. Experiments and Results
Figure 6 shows the experimental environment. In our experiments, Impinj R420 RFID reader and
EPC Gen2 tags are used. The experimental settings are listed in Table 2. In the experiments, the whole
5.1. Experimental Settings
test area is 3.6 m × 4.8 m. We placed 117 reference tags positioned in 9 rows by 13 columns in the
Figure 6 monitoring
shows the area of the RFID reader (Figure 6a shows part of the monitoring area) and 5 target tags
experimental environment. In our experiments, Impinj R420 RFID reader
which needed to be positioned. The matchboxes were used to bolster tags in the experiments. The
and EPC Gen2 tags are used. The with
detailed experimental settings experimental
every tag spacesettings are listed
duration distance in Table
are shown 2.6b,Inwhere
in Figure the experiments,
the squares and triangles represent reference tags and to-be-positioned target tags,
the whole test area is 3.6 m × 4.8 m. We placed 117 reference tags positioned in 9 rows by 13 columnsrespectively.

in the monitoring area of the RFID readerTable(Figure 6a shows

2. The experimental part of the monitoring area) and 5 target
settings. [Office1]: The Table 2. is revised as t
tags which needed to be positioned. The
Parameter
matchboxes were used to bolster
Value
tags in the experiments.
The detailed experimental settings with
Range every tag space duration 3.6distance
m × 4.8 m are shown in Figure 6b,
Numbers
where the squares and triangles of reference
represent tags
reference 117
tags and to-be-positioned target tags, respectively.
Numbers of target tags 5

(a)

ŏ ŏ [Office2]: This figure should be delet

 
 Figure 6. Cont. 
ŏ ŏ
45cm

ŏ 7DJ ŏ
7DJ
360cm ŏ 7DJ ŏ

ŏ 7DJ ŏ
7DJ
ŏ ŏ
 
 
ŏ ŏ
40cm 5HIHUHQFHWDJ 7DUJHWWDJ
480cm
Sensors 2017, 17, 1806 11 of 17
Sensors 2017, 17, 1806 11 of 17

Sensors 2017, 17, 1806 11 of 17

(b)
Figure 6. Experimental environment. (a) part of the monitoring area; (b) every tag space duration
distance.
Figure 6. Experimental environment. (a) part of the monitoring area; (b) every tag space duration distance.

5.2. Results of the Gaussian Filter for RSS

Table 2. The experimental settings.
We collected the 5 tags’ RSS values using the R420 RFID reader, Figure 7 shows the results of
the RSS values where we can see that there are many fluctuations. The Gaussian filter is used to
Parameter Value
process these abnormal data, Figure 8 shows the (b)
results of the filtered RSS value. We can see that the
RSS values processed by Range filter are stable to some3.6extent
the Gaussian m × 4.8
andm there are no large
Figure 6. Experimental environment. (a) part of the monitoring area; (b) every tag space duration
fluctuations. Numbers of reference tags 117
distance. Numbers of target tags 5

5.2. Results of the Gaussian Filter for RSS

5.2. Results of the Gaussian Filter for RSS
We collected the 5 tags’ RSS values using the R420 RFID reader, Figure 7 shows the results of
We
the collected
RSS valuesthe 5 tags’
where weRSS
can values
see thatusing
there the
are R420
manyRFID reader, The
fluctuations. Figure 7 shows
Gaussian theisresults
filter used toof the
RSS values where we can see that there are many fluctuations. The Gaussian filter is used
process these abnormal data, Figure 8 shows the results of the filtered RSS value. We can see that to the
process
theseRSS values data,
abnormal processed by 8the
Figure Gaussian
shows filter are
the results of stable to some
the filtered RSSextent and
value. Wethere
can are
see no
thatlarge
the RSS
fluctuations.
values processed by the Gaussian filter are stable to some extent and there are no large fluctuations.

Tag 1 Tag 2

Tag 1 Tag 2

Figure 7. Cont.
Sensors 2017, 17, 1806 12 of 17
Sensors 2017, 17, 1806 12 of 17

Tag 3 Tag 4

Tag 5

Figure 7. The RSS values of 5 tags.

Figure 7. The RSS values of 5 tags.

Tag 1 Tag 2
Tag 1 Tag 2

Tag 3 Tag 4

Figure 8. Cont.
 Sensors 2017, 17, 1806 13 of 17
Sensors 2017, 17, 1806 13 of 17
Tag 3 Tag 4

Tag 5

Figure 8. The RSS value of 5 tags processed by the Gaussian filter.

Figure 8. The RSS value of 5 tags processed by the Gaussian filter.
5.3. Results of the Average Location Error
5.3. ResultsFirstly,
of the Average Location
we selected Error data obtained from the RFID reader and used the LANDMARC
the original
algorithm to position unknown tags. Then the average location error was compared among the
Firstly, we selected the original data obtained from the RFID reader and used the LANDMARC
LANDMARC algorithm [25], the IKNN algorithm [43], the Gaussian-based algorithm and our BKNN
algorithm to position unknown tags. Then the average location error was compared among the
algorithm. In the experiments, 4 reference tags are selected to locate target. We performed 10 timed
LANDMARC algorithm [25], the IKNN algorithm [43], the Gaussian-based algorithm and our BKNN
tests, and each test took about 20 min. So the sum of Num-Announces of each selected reference tag
algorithm. In the
was about 8000, experiments, 4 reference
and all of those tagswere
RSS values are selected
collectedto locate
and usedtarget. We performed
by positioning algorithms10 timed
to
tests, and
estimate the 5 target tags. Figure 9 shows the results of the average location error among the tag
each test took about 20 min. So the sum of Num-Announces of each selected reference
was about 8000,algorithms.
compared and all ofWe those RSSthat
can see values were collected
the location error of theand used by positioning
Gaussian-based algorithm is algorithms
less than to
thatthe
estimate of 5the LANDMARC
target tags. Figure and9 IKNN
showsalgorithm,
the resultsthis is because
of the averagesome abnormal
location error RSS
among values
the can be
compared
filtered by the Gaussian filter. The average error of the BKNN location
algorithms. We can see that the location error of the Gaussian-based algorithm is less than that ofestimation is about 15 cm,
which is the smallest
the LANDMARC and IKNN among the compared
algorithm, thisalgorithms.
is becauseThis some is because
abnormalour RSS
BKNN uses the
values canproper k
be filtered
value and processed RSS values which can remove abnormal location data, and uses Bayesian
by the Gaussian filter. The average error of the BKNN location estimation is about 15 cm, which is
probability to estimate the target tag’s location which can predict the exact location data. Thus, BKNN
the smallest among the compared algorithms. This is because our BKNN uses the proper k value
reduces the location error and has a much higher location accuracy than that of other compared
and processed RSS values which can remove abnormal location data, and uses Bayesian probability
algorithms. According to literature [43], the tag space between the two reference tags is 1 m, but the
to estimate the
average error target tag’s location
of location estimationwhich can85
is about predict
cm, whilethethe
exact location
average errordata.
of ourThus, BKNN
algorithm reduces
is about
the location error the
15 cm when andtag has a much
space is 40 higher location
cm. Though accuracy
we use than that
more reference of other
tags, compared
the accuracy algorithms.
is improved,
According
whichtohas literature [43],implications
significant the tag space for between
practical the twopositioning
indoor reference tags is 1 m, but
algorithms. In lowthecost
average
RFID-error
of location estimation is about 85 cm, while the average error of our algorithm is about 15 cmour
based indoor positioning environments, the average accuracy of the location estimation of when
the tagalgorithm
space isis40also cm.very effective
Though weandusesuitable for use intags,
more reference indoorthelocation
accuracy based services, i.e.,
is improved, book has
which
searching
significant in the library,
implications for locating
practical patients
indoorinpositioning
a hospital, finding people trapped
algorithms. In low costin a burning
RFID-based building,
indoor
finding an individual’s luggage without removing it from the vehicle or the aircraft in an airport and
positioning environments, the average accuracy of the location estimation of our algorithm is also
reminding passengers of flights and boarding gates so that they do not miss their flights or delay
very effective and suitable for use in indoor location based services, i.e., book searching in the library,
departures.
locating patients in a hospital, finding people trapped in a burning building, finding an individual’s
luggage without removing it from the vehicle or the aircraft in an airport and reminding passengers of
flights and boarding gates so that they do not miss their flights or delay departures.
 Sensors2017,
Sensors 2017,17,
17,1806
1806 1414ofof1717

Figure9.9.The
Figure Theaverage
averageerror
errorofoflocation
locationestimation
estimationofof5 5tags.
tags.

5.4.
5.4.Comparisons
ComparisonsofofBKNN
BKNNwith
withDifferent
DifferentPositioning
PositioningMethods
Methods
InInthis
thissection, forfor
section, precision, cost,cost,
precision, timetime
consumption,
consumption,energyenergy
consumption, etc., weetc.,
consumption, compare BKNN
we compare
with
BKNN other positioning
with methods, methods,
other positioning such as LANDMARC
such as LANDMARC[25], fingerprint-based [33], phase-based
[25], fingerprint-based [34]
[33], phase-
and IKNN
based [34][43].
andTable
IKNN3 shows the results
[43]. Table 3 showsof the comparisons.
results of the As passive UHF
comparisons. AsRFID wasUHF
passive used,RFID
the cost
was
and energy
used, consumption
the cost and energyisconsumption
low in our BKNN is lowalgorithm.
in our BKNN All the compared
algorithm. Allalgorithms
the compared use algorithms
the COTS
RFID reader
use the COTS which
RFIDcanreader
easily which
be implemented
can easilyin becurrent RFID systems,
implemented and RFID
in current because the abnormal
systems, data
and because
isthe
filtered
abnormalbefore calculating
data is filteredthe finalcalculating
before location, the thetime
finalconsumption
location, the is lower
time than that of
consumption is the IKNN
lower than
and
thatLANDMARC
of the IKNN and algorithms.
LANDMARC algorithms.

Table3.3.Comparisons
Table ComparisonsofofBKNN
BKNNwith
withdifferent
differentpositioning
positioningmethods.
methods.

Index Index Time Energy COTS

Precision Time Cost Energy
Consumption
COTS
Positioning
Positioning Methods Precision Consumption Cost Reader
Consumption Consumption Reader
Methods
LANDMARC [25] Normal High Normal Normal Yes
Fingerprint [33]
LANDMARC [25] High Normal Low High High NormalLow NormalYes Yes
Phase [34] High High Normal Normal Yes
IKNNFingerprint
[43] [33]
High High High Low Normal HighNormal Low Yes Yes
Proposed BKNNPhase [34] High High Normal High Low NormalLow NormalYes Yes
IKNN [43] High High Normal Normal Yes
6. Conclusions BKNN
Proposed High Normal Low Low Yes

This paper has presented an indoor positioning method using passive UHF RFID. However,
6. Conclusions
there are two problems that passive RFID encounters when used in an indoor location. The first
This
problem is paper
that thehasRSSpresented
value ofan indoor
RFID positioning
is easily affectedmethod using passivedynamic
by the surrounding UHF RFID. However,
environment.
The second problem is the phase value is cyclical and the phase difference is tag sensitive,The
there are two problems that passive RFID encounters when used in an indoor location. first
which
problem
makes UHF is that
RFIDthe RSS value
difficult of in
to use RFID is easilylocations.
large-scale affected byInthethissurrounding dynamic
paper, we mainly environment.
solve the noise
problem which influences the location accuracy. A Gaussian filter is used to filter abnormalwhich
The second problem is the phase value is cyclical and the phase difference is tag sensitive, RSS
makes UHF RFID difficult to use in large-scale locations. In this paper,
values and Bayesian estimation together with the KNN algorithm are used to improve positioningwe mainly solve the noise
problem In
accuracy. which influences the
the experiments, location tags
4 reference accuracy. A Gaussian
are selected to locatefilter is used
target. to filter location
The average abnormal RSS
error
values and Bayesian estimation together with the KNN algorithm are used
of our algorithm is about 15 cm, which is the smallest among the compared algorithms. Because theto improve positioning
accuracy. In
positioning the experiments,
accuracy 4 reference
has a relationship withtags
theare selected
collected to values
RSS locate target. The average
of reference location
tags, how error
to deploy
of our algorithm is about 15 cm, which is the smallest among the compared
the optimized placement of readers and reference tags will be studied in our future work. algorithms. Because the
positioning accuracy has a relationship with the collected RSS values of reference tags, how to deploy
the optimized placement of readers and reference tags will be studied in our future work.
Sensors 2017, 17, 1806 15 of 17

Acknowledgments: The authors thank all the reviewers and editors for their valuable comments and work.
This work is financially supported by Jiangsu Government Scholarship for Overseas Studies, the National Natural
Science Foundation of P. R. China (No.: 61373017, No.: 61572260, No.: 61572261, No.: 61672296, No.: 61602261),
the Natural Science Foundation of Jiangsu Province (No.: BK20140886, No.: BK20140888), Jiangsu Natural
Science Foundation for Excellent Young Scholars (BK20160089), Scientific & Technological Support Project of
Jiangsu Province (No.: BE2015702, BE2016185, No.: BE2016777), China Postdoctoral Science Foundation (No.:
2014M551636, No.: 2014M561696), Jiangsu Planned Projects for Postdoctoral Research Funds (No.: 1302090B,
No.: 1401005B), Postgraduate Research and Practice Innovation Program of Jiangsu Province (SJLX16_0326,
KYCX17_0798), Project of Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks
(WSNLBZY201509) and NUPTSF (Grant No.: NY214060, No. NY214061).
Author Contributions: He Xu and Ye Ding conceived and designed the experiments, and wrote the paper; Peng Li
performed the experiments and analyzed the data; Ruchuan Wang contributed RFID devices and guided their use.
Yizhu Li helped with language modification. All authors contributed to the final version.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Ijaz, F.; Yang, H.K.; Ahmad, A.W.; Lee, C. Indoor positioning: A review of indoor ultrasonic positioning
systems. In Proceedings of the 15th International Conference on Advanced Communication Technology
(ICACT), PyeongChang, Korea, 27–30 January 2013; pp. 1146–1150.
2. Gupta, A.; Yilmaz, A. Indoor Positioning using Visual and Inertial Sensors. In Proceedings of the IEEE
Sensors, Orlando, FL, USA, 30 October–3 November 2016; pp. 1–3.
3. Bekkelien, A. Bluetooth Indoor Positioning; University of Geneva: Geneva, Switzerland, 2012.
4. Yang, C.; Shao, H.R. WiFi-based indoor positioning. IEEE Commun. Mag. 2015, 53, 150–157. [CrossRef]
5. Blankenbach, J.; Norrdine, A.; Hellmers, H. A robust and precise 3D indoor positioning system for harsh
environments. In Proceedings of the 2012 International Conference on Indoor Positioning and Indoor
Navigation (IPIN), Sydney, Australia, 13–15 November 2012; pp. 1–8.
6. Yayan, U.; Yucel, H.; Yazici, A. A low cost ultrasonic based positioning system for the indoor navigation of
mobile robots. J. Intell. Robot. Syst. 2015, 78, 541–552. [CrossRef]
7. Deng, Z.-A.; Wang, G.F.; Hu, Y.; Cui, Y. Carrying Position Independent User Heading Estimation for Indoor
Pedestrian Navigation with Smartphones. Sensors 2016, 16, 677. [CrossRef] [PubMed]
8. Li, H. Low-cost 3d bluetooth indoor positioning with least square. Wirel. Pers. Commun. 2014, 78, 1331–1344.
[CrossRef]
9. Kim, H.S.; Seo, W.; Baek, K.R. Indoor positioning system using magnetic field map navigation and an encoder
system. Sensors 2017, 17, 651. [CrossRef] [PubMed]
10. Nascimento, H.J.; Rodrigues, E.B.; Cavalcanti, F.R.; Paiva, A.R.L. An Algorithm Based on Bayes Inference
And K-nearest Neighbor for 3D WLAN Indoor Positioning. In Proceedings of the XXXIV Simposio Brasileiro
De Telecomunicacoes (SBRT), Santarem, Brasil, 30 August–2 September 2016; pp. 398–402.
11. Chao, C.T.; Chung, M.H.; Chiou, J.S.; Wang, C.J. A simple interface for 3d position estimation of a mobile
robot with single camera. Sensors 2016, 16, 435. [CrossRef] [PubMed]
12. Finkenzeller, K. RFID Handbook: Fundamentals and Applications in Contactless Smart Cards, Radio Frequency
Identification and Near-Field Communication; John Wiley & Sons: Hoboken, NJ, USA, 2010.
13. Zhang, J.; Tian, G.Y.; Marindra, A.M.; Sunny, A.I.; Zhao, A.B. A review of passive RFID tag antenna-based
sensors and systems for structural health monitoring applications. Sensors 2017, 17, 265. [CrossRef] [PubMed]
14. Martínez Pérez, M.; Vázquez González, G.; Dafonte, C. Evaluation of a Tracking System for Patients and
Mixed Intravenous Medication Based on RFID Technology. Sensors 2016, 16, 2031. [CrossRef] [PubMed]
15. Kim, S.C.; Jeong, Y.S.; Park, S.O. RFID-based indoor location tracking to ensure the safety of the elderly
in smart home environments. Pers. Ubiquitous Comput. 2013, 17, 1699–1707. [CrossRef]
16. Mora-Mora, H.; Gilart-Iglesias, V.; Gil, D.; Sirvent-Llamas, A. A computational architecture based on RFID
sensors for traceability in smart cities. Sensors 2015, 15, 13591–13626. [CrossRef] [PubMed]
17. Prinsloo, J.; Malekian, R. Accurate vehicle location system using RFID, an internet of things approach.
Sensors 2016, 16, 825. [CrossRef] [PubMed]
18. Valero, E.; Adán, A.; Cerrada, C. Evolution of RFID applications in construction: A literature review. Sensors
2015, 15, 15988–16008. [CrossRef] [PubMed]
Sensors 2017, 17, 1806 16 of 17

19. Fan, H.; Wu, Q.; Lin, Y.; Zhang, J. A split-path schema-based rfid data storage model in supply chain
management. Sensors 2013, 13, 5757–5776. [CrossRef] [PubMed]
20. Chai, J.; Wu, C.; Zhao, C.; Chi, H.L.; Wang, X.; Ling, B.W.K.; Teo, K.L. Reference tag supported RFID tracking
using robust support vector regression and Kalman filter. Adv. Eng. Inform. 2017, 32, 1–10. [CrossRef]
21. Song, X.; Li, X.; Tang, W.; Zhang, W.; Li, B. A hybrid positioning strategy for vehicles in a tunnel based on
RFID and in-vehicle sensors. Sensors 2014, 14, 23095–23118. [CrossRef] [PubMed]
22. Shen, B.; Zheng, Q.; Li, X.; Xu, L. A framework for mining actionable navigation patterns from in-store RFID
datasets via indoor mapping. Sensors 2015, 15, 5344–5375. [CrossRef] [PubMed]
23. Hsu, C.C.; Chen, J.H. A novel sensor-assisted RFID-based indoor tracking system for the elderly living alone.
Sensors 2011, 11, 10094–10113. [CrossRef] [PubMed]
24. Domdouzis, K.; Kumar, B.; Anumba, C. Radio-Frequency Identification (RFID) applications: A brief
introduction. Adv. Eng. Inform. 2007, 21, 350–355. [CrossRef]
25. Ni, L.M.; Liu, Y.; Lau, Y.C.; Patil, A.P. LANDMARC: Indoor location sensing using active RFID. Wirel. Netw.
2004, 10, 701–710. [CrossRef]
26. Han, S.; Kim, J.; Park, C.H.; Yoon, H.C.; Heo, J. Optimal detection range of RFID tag for RFID-based
positioning system using the k-NN algorithm. Sensors 2009, 9, 4543–4558. [CrossRef] [PubMed]
27. Benelli, G.; Bertoni, D.; Sarti, G. An analysis on the use of LF RFID for the tracking of different typologies of
pebbles on beaches. In Proceedings of the 2011 IEEE International Conference on RFID-Technologies and
Applications (RFID-TA), Sitges, Spain, 15–16 September 2011; pp. 426–431.
28. Mi, J.; Takahashi, Y. Design of an HF-Band RFID System with Multiple Readers and Passive Tags for Indoor
Mobile Robot Self-Localization. Sensors 2016, 16, 1200. [CrossRef] [PubMed]
29. Own, C.M.; Lee, D.S.; Wang, T.H.; Wang, D.J.; Ting, Y.L. Performance Evaluation of UHF RFID Technologies
for Real-Time Bus Recognition in the Taipei Bus Station. Sensors 2013, 13, 7797–7812. [CrossRef] [PubMed]
30. Xu, C.; Law, C.L. TOA estimator for UWB backscattering RFID system with clutter suppression capability.
EURASIP J. Wirel. Commun. Netw. 2010, 753129. [CrossRef]
31. Zhou, J.; Zhang, H.; Mo, L. Two-dimension localization of passive RFID tags using AOA estimation.
In Proceedings of the 2011 IEEE Instrumentation and Measurement Technology Conference (I2MTC),
Hangzhou, China, 10–12 May 2011; pp. 1–5.
32. Bouchard, K.; Fortin-Simard, D.; Gaboury, S.; Bouchard, B.; Bouzouane, A. Accurate trilateration for passive
RFID localization in smart homes. Int. J. Wirel. Inf. Netw. 2014, 21, 32–47. [CrossRef]
33. Soonjun, S.; Boontri, D.; Cherntanomwong, P. A novel approach of RFID based indoor localization using
fingerprinting techniques. In Proceedings of the 15th Asia-Pacific Conference on Communications (APCC),
Shanghai, China, 8–10 October 2009; pp. 475–478.
34. Buffi, A.; Nepa, P.; Lombardini, F. A phase-based technique for localization of UHF-RFID tags moving on a
conveyor belt: Performance analysis and test-case measurements. IEEE Sens. J. 2015, 15, 387–396. [CrossRef]
35. Guidi, F.; Sibille, A.; Roblin, C.; Casadei, V.; Dardari, D. Analysis of UWB tag backscattering and its impact
on the detection coverage. IEEE Trans. Antennas Propag. 2014, 62, 4292–4303. [CrossRef]
36. Guidi, F.; Decarli, N.; Bartoletti, S.; Conti, A.; Dardari, D. Detection of multiple tags based on impulsive
backscattered signals. IEEE Trans. Commun. 2014, 62, 3918–3930. [CrossRef]
37. Arnitz, D.; Muehlmann, U.; Witrisal, K. UWB ranging in passive UHF RFID: Proof of concept. Electron. Lett.
2010, 46, 1401–1402. [CrossRef]
38. Zhao, C.; Wu, C.; Chai, J.; Wang, X.; Yang, X.; Lee, J.M.; Kim, M.J. Decomposition-based multi-objective firefly
algorithm for RFID network planning with uncertainty. Appl. Soft Comput. 2017, 55, 549–564. [CrossRef]
39. Guidi, F.; Decarli, N.; Dardari, D.; Mani, F.; D'Errico, R. Passive Millimeter-Wave RFID using Backscattered
Signals. In Proceedings of the 2016 IEEE Globecom Workshops (GC Wkshps), Washington, DC, USA,
4–8 December 2016; pp. 1–6.
40. Li, D.; Zhang, B.; Li, C. A Feature-Scaling-Based k-Nearest Neighbor Algorithm for Indoor Positioning
Systems. IEEE Internet Things J. 2016, 3, 590–597. [CrossRef]
41. Liang, X.; Gou, X.; Liu, Y. Fingerprint-based location positioning using improved KNN. In Proceedings of
the 3rd IEEE International Conference on Network Infrastructure and Digital Content (IC-NIDC), Beijing,
China, 21–23 September 2012; pp. 57–61.
Sensors 2017, 17, 1806 17 of 17

42. Han, K.; Cho, S.H. Advanced LANDMARC with adaptive k-nearest algorithm for RFID location system.
In Proceedings of the 2nd IEEE International Conference on Network Infrastructure and Digital Content,
Beijing, China, 24–26 September 2010; pp. 595–598.
43. Liu, X.; Wen, M.; Qin, G.; Liu, R. LANDMARC with improved k-nearest algorithm for RFID location
system. In Proceedings of the 2nd IEEE International Conference on Computer and Communications (ICCC),
Chengdu, China, 14–17 October 2016; pp. 2569–2572.
44. Sun, Y.; Gu, F. Compressive sensing of piezoelectric sensor response signal for phased array structural health
monitoring. Int. J. Sens. Netw. 2017, 23, 258–264. [CrossRef]
45. Nikitin, P.V.; Martinez, R.; Ramamurthy, S.; Leland, H.; Spiess, G.; Rao, K.V.S. Phase based spatial
identification of UHF RFID tags. In Proceedings of the 2010 IEEE International Conference on RFID,
Orlando, FL, USA, 14–16 April 2010; pp. 102–109.
46. Catarinucci, L.; Colella, R.; Tarricone, L. Design, development, and performance evaluation of a compact
and long-range passive UHF RFID tag. Microw. Opt. Technol. Lett. 2012, 54, 1335–1339. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).