Professional Documents
Culture Documents
Performance Analysis of TDOA-based Indoor Positioning Systems Using Visible LED Lights
Performance Analysis of TDOA-based Indoor Positioning Systems Using Visible LED Lights
Performance Analysis of TDOA-based Indoor Positioning Systems Using Visible LED Lights
I. I NTRODUCTION
Over the last decade, the rapid proliferation of wireless readily available inside buildings. As shown in Figure 1,
devices (smart phones, smart watches) has led to the in such designs, the ceiling LED lights are used as
development of ancillary Internet of Things (IoT) ser- transmitters, and a light sensor (such as, a photodiode
vices, such as, Location Based Services (LBS) in indoor or an image sensor) is used as a receiver whose location
environments. Indoor LBS applications can benefit a wide coordinates are required to be determined [7].
range of consumers for surveillance, health monitoring, VLC-IPS designs can exploit the variations in Received
disaster management, asset monitoring, robot navigation, Signal Strength (RSS), Time of Arrival (TOA) and Time
building management, etc., [1]–[6]. However, successful Difference of Arrival (TDOA) of the signals that are
LBS application development requires the ability to de- transmitted by various LEDs [8]. However, there are
termine the location coordinates of a receiver with high certain practical limitations on the use of RSS and TOA in
accuracy and low cost. such systems. For instance, the performance of RSS-based
Among the available options for position determina- VLC-IPS is degraded if the receiver lies in areas where the
tion in indoor environment, Visible Light Communication available light intensity is significantly poor. Moreover,
(VLC) based Indoor Positioning System (IPS) designs, the reflection of signals may also have adverse effects
VLC-IPS, have the advantage of leveraging the exist- on RSS that can reduce the localization accuracy [9].
ing Light Emitting Diode (LED) lighting infrastructure Similarly, TOA-based VLC-IPS require perfect synchro-
This research is supported by Ignite, formerly known as National ICT nization between LEDs and receiver to have accurate
R&D Fund, Pakistan, through grant number SRG-129. estimates of time of arrival of signals from different LED
5m
receiver is given by,
N
y(t) = G si (t) ⊗ hi (t) + n(t) (2)
i=1
LED-3
where G is the responsiveness of the PD, hi (t) is the
optical channel impulse response and n(t) is additive
Figure 2. LED arrangement on ceiling of the room
white Gaussian noise (AWGN).
As shown in Figure 3, a bandpass filter (BPF) bank
sources. On the other hand, TDOA-based VLC-IPS do not separates out the individual LED signals, denoted by
depend on the received light intensity and therefore can vi (t), which in terms of time difference of arrival (τi )
provide relatively better performance even in poorly lit are given by [10],
areas [8]. Consequently, TDOA-based VLC-IPS designs
have the potential to provide extremely low localization vi (t) = K̃i cos (2πfi (t − τi )) + n(t) (3)
errors that are in the range of a few centimeters [10], [11].
One primary tool to establish and compare the per- where K̃i captures the effect of signal attenuation, τi =
di th
formance of various systems is through the determina- c , di is the distance between i LED and the receiver
tion of the Cramer Rao Lower Bound (CRLB) [12]. and c is the speed of light. Let ri = (xi , yi ) denote
For RSS- and TOA-based VLC-IPS, existing literature the location coordinates of the ith LED. Similarly, let
provides CRLB [13]–[16]. However, such analysis is not r = (x, y) denote the unknown location coordinates of the
yet available for TDOA-based VLC-IPS. In this paper, receiver. The Euclidean distance
between the receiver and
we are bridging this gap and derive the CRLB on the the ith LED is equal to di = (x − xi )2 + (y − yi )2 .
localization accuracy of TDOA-based VLC-IPS designs. For obtaining TDOA measurements, LED-1 is taken
We also identify the conditions and study the impact of as a reference and it is paired with each of the re-
different factors that minimize the bound and could result maining LEDs to get different measurement sets. Each
in maximization of localization accuracy. signal is passed through the frequency down converter
The rest of the paper is organized as follows: Section II block, which multiplies vi (t) with cos 2π fi − f21 t
describes the system model that is used to evaluate the followed by a low pass filter. The phase difference be-
performance of TDOA-based VLC-IPS. In Section III, tween the resulting signal and that of LED-1 is measured
CRLB is derived for TDOA-based systems. Section IV (e.g., using Hilbert transform method). Using the phase
provides the simulation results and discusses the factors difference, noisy estimates of {Δdi,1 = di − d1 }N i=2 can
that affect the localization accuracy. The paper is finally be obtained (the derivation is given in [10]). Let,
concluded in Section V.
II. S YSTEM M ODEL zi,1 = Δdi,1 + ni,1 , i = 2, . . . , N (4)
In this section, we describe the system model that is where, ni,1 is the noise in the estimation of Δdi,1 .
used to analyze the localization accuracy. We assume that As in [17], we assume ni,1 to be zero mean additive
there are N LEDs installed on the ceiling. Each LED is 2
white Gaussian noise with variance σi,1 . Let, Δd =
transmitting with an average power Pt . Figure 2 shows [Δd2,1 Δd3,1 . . . ΔdN,1 ]T and z = [z2,1 z3,1 . . . zN,1 ]T (T
the LED arrangement on the ceiling, while the overall denotes the transpose operation). Using Δd, the unknown
placement of the transmitter and receiver is shown in location coordinates of the receiver can be determined as
Figure 1 where 1 to 5 are the five LEDs. We assume explained in [10]. However, in actual practice, the location
Frequency Division Multiplexing (FDM) where each LED has to be estimated based on the noisy estimates z that
is assigned a different modulation frequency, fi , at which impacts the performance.
it transmits its location coordinates. For TDOA measure-
ments, LED-1 is designated as a reference LED with a III. C OMPUTATION OF THE CRLB
modulation frequency f1 . The transmission frequencies of The derivation of the CRLB for TDOA-based VLC-
all other LEDs are assumed to be odd multiples of f1 [10]. IPS requires the computation of the inverse of Fisher
104
Transmitter Channel Receiver
n(t)
Figure 3. Block diagram of the overall communication having the channel noise, the transmitter and the receiver
105
is obvious that θi , which is the incidence angle is equal
to the transmission angle, φi . Moreover, cos(φi ) = h/di
where h is the vertical distance of the LEDs from the
receiver. Thus, we obtain,
G2 A2r h4 Pt2 m
γi = = 8 (11)
π 2 d8i N0 di
where m = (G2 A2r h4 Pt2 )/(π 2 N0 ). Since LEDs trans-
mit at frequencies which are odd multiples of reference
2 a2 d8i d8j
σi,j ≥ + (12)
(f1 )2 (2i − 1)2 (2j − 1)2
where a2 = (3c2 )/(8π 2 m). Once the variances are
known, Σ is also determined.
IV. S IMULATION R ESULTS AND D ISCUSSION
We assume a simulation setup of a 5 m×5 m×3 m room
with five LEDs installed in an arrangement as is shown in
Figure 2. The reference LED is placed in the center of the
106
[7] Brena, R.F., Garcı́a-Vázquez, J.P., Galván-Tejada, C.E., Muñoz-
3 LEDs
Distance dependent variance [m2]
5 LEDs
Rodriguez, D., Vargas-Rosales, C. and Fangmeyer, J., “Evolution
0.25 7 LEDs of Indoor Positioning Technologies: A Survey”, Journal of Sen-
sors, 2017.
[8] Trong-Hop, D. and Myungsik, Y.:, “An in-Depth Survey of Visible
0.2 Light Communication Based Positioning Systems,” Sensors, 2016,
vol. 16, no. 5, 2016.
[9] Wu, Nan, Lihui Feng, and Aiying Yang, “Localization Accuracy
0.15 Improvement of a Visible Light Positioning System Based on the
Linear Illumination of LED Sources,” IEEE Photonics Journal,
vol. 9, no. 5, pp. 1-11, 2017.
0.1 [10] Nadeem, U. and Hassan, N. U. and Pasha, M. A. and Yuen, C.:, “A
Highly Accurate Three Dimensional Wireless Indoor Positioning
System Using White LED Lights,” Electronics Letters, vol. 50, no.
0.05
11, pp. 828-830, 2014.
[11] Naz, A., Hassan, N. U., Pasha M. A., Asif H., Jadoon T., and
0 Yuen, C.:, “Single LED Ceiling Lamp Based Indoor Positioning
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 System”, IEEE 4th World Forum on Internet of Things (WF-IoT),
Frequency of the reference LED (MHz) 2018, 5-8 Feb. 2018
[12] Steendam, H. and Wang T.Q. and Armstrong, J.:, “Theoretical
Lower Bound for Indoor Visible Light Positioning Using Received
Figure 7. Dependence of variance on the frequency of Signal Strength Measurements and an Aperture-Based Receiver,”
reference LED and the number of LEDs Lightwave Technology, vol. 35, no. 2, pp. 309-319, 2017.
[13] H. Steendam, T. Wang, and J. Armstrong, “Theoretical Lower
Bound for Indoor Visible Light Positioning Using Received Signal
Strength Measurements and an Aperture-Based Receiver”, Light-
that the accuracy performance obtained through TDOA- wave Technology, vol. 35, pp. 309-319, 2017.
based VLC-IPS is comparable to that obtained through [14] Wang, T.Q. and Sekercioglu, Y.A. and Neild, A. and Armstrong,
J.:, “Position Accuracy of Time-of-Arrival Based Ranging Using
RSS [12]. However, unlike latter, the performance of the Visible Light With Application in Indoor Localization Systems,”
former is not degraded by reduced signal strength of the Lightwave Technology, vol. 31, no. 20, pp. 3302-3308, 2013.
incoming light signals, which makes the former an apt [15] Zhang, X. and Duan, J. and Fu, Y. and Shi, A.:, “Theoretical Accu-
racy Analysis of Indoor Visible Light Communication Positioning
choice for highly accurate VLC-IPS. System Based on Received Signal Strength Indicator,” Lightwave
Technology, vol. 32, no. 21, 2014.
[16] M. F. Keskin and S. Gezici, “Comparative Theoretical Analysis
V. C ONCLUSION of Distance Estimation in Visible Light Positioning Systems,”
Lightwave Technology, vol. 34, no. 3, pp. 854-865, 2016.
We have analyzed the theoretical accuracy of a TDOA- [17] Taparugssanagorn, A. and Siwamogsatham, S. and Pomalaza-Raez,
based VLC-IPS design by computing the CRLB. We C.:, “A Hexagonal Coverage LED-ID Indoor Positioning Based on
also identified the factors that influence this bound. The TDOA with Extended Kalman Filter,” IEEE Computer Software
and Applications Conference, pp. 724-747, 2013.
results show that with 5 LEDs and a reference frequency [18] Yang, H., Bergmans, J., and Schenk, T.:, “Illumination sensing in
of 1 MHz, TDOA-based systems can localize a receiver LED lighting systems based on frequency-division multiplexing”,
within an accuracy of few centimeters, making it an apt IEEE Trans. Signal Process., 2009, 57, (11), pp. 4269–4281
[19] Zhou, Z., Kavehrad, M., and Deng, P.:, “Indoor positioning al-
choice for implementation in accurate indoor localization gorithm using light-emitting diode visible light communications”,
applications. Opt. Eng., 2012, 51, (8), pp. 085009–(1–6)
[20] Jung, S.Y., Hann, S., and Park, C.S.:, “TDOA-based Optical
Wireless Indoor Localization using LED Ceiling Lamps”, IEEE
R EFERENCES Trans. Consum. Electron., 2011, 57, (4), pp. 1592–1597
[1] A. Petcovici and E. Stroulia, “Location-based services on a smart
campus: A system and a study”, IEEE 3rd World Forum on Internet
of Things (WF-IoT), pp. 94-99, 2016.
[2] M. Amarlingam, C. N. Firoz and P. Rajalakshmi, “Mobile phone
based acoustic localization using Doppler shift for wireless sensor
networks”, IEEE 3rd World Forum on Internet of Things (WF-IoT),
pp. 283-288, 2016.
[3] N. U. Hassan, A. Naeem, M. A. Pasha, T. Jadoon, and C. Yuen.
“Indoor Positioning Using Visible LED Lights: A Survey,” ACM
Comput. Surv., vol. 48, no. 2, pp. 20:1-23:2, 2015.
[4] K. Qiu, F. Zhang and M. Liu, “Visible Light Communication-based
indoor localization using Gaussian Process,” IEEE International
Conference on Intelligent Robots and Systems (IROS), pp. 3125-
3130, 2015.
[5] J. Yan and B. Zhu, “A visible light communication indoor localiza-
tion algorithm in rotated environments,” International Conference
on Computer, Information and Telecommunication Systems (CITS),
pp. 1-4, 2016.
[6] Yi, Keon Young, Dae Young Kim, and Kwang Moo Yi, “De-
velopment of a Localization System Based on VLC Technique
for an Indoor Environment,” Journal Of Electrical Engineering &
Technology, vol. 10, no. 1, pp. 436-442, 2015.
107