Sie sind auf Seite 1von 8

Photon Netw Commun

DOI 10.1007/s11107-015-0507-1

Indoor visible light communication without line of sight:


investigation and performance analysis
Maged A. Esmail1,2

Habib A. Fathallah1,2

Received: 19 January 2015 / Accepted: 15 April 2015


Springer Science+Business Media New York 2015

Abstract Visible light communication (VLC) using indoor


LED lighting generally assumes the existence of line of sight
link in addition to multipath, delayed, lower power reflections. In this paper, we investigate the possibility to establish
VLC links in shadowed areas, i.e., where the line of sight
is blocked or unavailable. First, we study the system performance in terms of received power, SNR, BER, and rms delay
spread. The results show acceptable, yet promising performance of BER = 1.2 103 in worst non-line of sight
case. Second, we define three configuration scenarios for
receivers PD orientation in which line of sight is absent.
For each scenario, we evaluate the link performance. Our
analysis shows that the system performance can be potentially improved when MIMO is considered for locations that
have low performance with single photodetector.
Keywords Visible light communication Optical
communication Shadowing Non-line of sight LED
Reflection

1 Introduction
During the last decade, visible light communication (VLC)
observed an increasing importance in various research circles [1,2]. The interest in this new technology is motivated
by the rapid development in LED technology. Compared to
other lighting sources, LEDs have a number of advantages.

Maged A. Esmail
mesmail@ksu.edu.sa

Electrical Engineering Department, King Saud University,


P.O. Box 800, Riyadh 11421, Saudi Arabia

KACST Technology Innovation Center in Radiofrequency


and Photonics for the e-Society (RFTONICS), King Saud
University, P.O. Box 800, Riyadh 11421, Saudi Arabia

For example, longer life expectancy, mercury free, smaller


size, high energy efficiency, low power consumption, etc. [3].
By the end of this decade, it is expected that LED will dominate the lighting market [4]. LED is a semiconductor, and
it can be modulated (switched ON and OFF) at high speed.
This key specification motivated many researchers to exploit
LEDs in data communication with very limited additional
cost. In addition to the use of LEDs for lighting and indoor
communications, LEDs have found many new applications
that appeared during the last few years. This includes indoor
(2D and 3D) positioning and navigation systems, advertising,
vehicle-to-vehicle communication, underwater communication, aircraft and hospital communication, etc. [5].
Many studies have been conducted on indoor VLC system
to characterize its channel and improve its performance. Most
of these studies have assumed the existence of the line of sight
(LOS) between the transmitter and the receiver. Furthermore,
the link performance was always evaluated mainly based on
the power carried by the LOS, i.e., reflected light paths are
usually neglected [1,3,5]. However, in reality, the LOS signal can be blocked permanently or temporally by obstacles
between the LED and the photodetector (PD). Therefore, the
non-LOS (NLOS) link that is produced by light reflections
can be used to maintain communication. In [6], the authors
investigated the performance of the NLOS communication
system by measuring the received power at few receivers
locations in the room. Then they calculated theoretically the
SNR, interference, and BER at specific values of the measured received power.
In this paper, we investigate the possibility of establishing
VLC when the LOS signal is blocked, relying on reflected
lights from the walls. For this purpose, we exploited commercial LED data to produce reliable results. The system
performance is evaluated using different metrics such as
the received power strength, SNR, interference, BER, and

123

Photon Netw Commun

2 VLC system model


The transmitted signal in indoor wireless channel encounters
reflections from walls and objects. This causes time dispersion and power fluctuation in the received signal. Specular
reflection can occur from mirrors and other shiny objects.
However, most reflections are diffuse in nature and most are
well modeled as Lambertian [7].
2.1 Illuminance
Since the preliminary objective for LED installation is lighting, the VLC should not affect the illumination. The luminous
intensity generated by an LED that has a Lambertian radiation pattern is given by
I () = I (0) cosm ()

(1)

where I (0) is the center luminous intensity of the LED and


is the angle of irradiance [8]. The order of Lambertian emission, m = 1/ log2 (cos 1/2 ), is a measure of the directivity
of the light beam and is related directly to semi-angle at half
power (1/2 ) of the LED. The horizontal illuminance at a
point (x, y) is given by
E hor (x, y) = I () cos( )/d 2

123

(2)

-3

x10

Plaster
()
Plasc

4.8

0.8

3.6

0.6

2.4

0.4

1.2

0.2

400

450

500

550

600

650

700

750

Spectral Reectance

Power Spectral Distribuon

transmitted data rate. These metrics are calculated not at specific locations, but in every point in the room at receiver
height to determine the limitations of the system performance. Moreover, we investigate the performance of the
system when increasing the number of detectors in the
receiver (MIMO system). Three different receiver configurations are used and compared. For each configuration, we
calculated the rms delay spread, SNR, BER, and data rate.
The results show improvement in the system performance
when MIMO system is used.
It is worthy to note that NLOS VLC could serve as an
excellent backup link (even with reduced capacity and performance) when the LOS is interrupted. This helps to maintain
continuous communication between the end user mobile and
the emitting LEDs. Moreover, it can be easily noted that
indoor building designers frequently fix the lighting diodes
in hidden corners and parts of the rooms. In that case, no
LOS may exist.
The remaining of the paper is organized as follows. In
Sect. 2, we study the system model including the channel
characteristics when no LOS is available. In Sect. 3, we
evaluate the system performance in terms of SNR, BER,
and transmitted data rate. We also show how the performance improves when the number of detectors is increased
in Sect. 4. Finally, we conclude in Sect. 5.

Wavelength (nm)

Fig. 1 PSD (left sidesolid line), and spectral reflectance for different
materials (right sidedashed lines)

where is the incidence angle and d is the distance between


the transmitter and the illuminated surface [8].
2.2 Channel impulse response
The light produced by the LED for lighting purpose covers the visible band (380780 nm) and other wavelengths.
Therefore, data transmission by LED covers wide range of
wavelengths. This is not the case with IR where single wavelength is used for data transmission. Therefore, the radiated
power in VLC is wavelength dependent and distributed over
the visible band. This dependency is described by the power
spectral distribution (PSD) () which describes the radiant power per unit wavelength. Moreover, the reflectivity
of objects in indoor environment is a function of the signal wavelength. Figure 1 shows the PSD for high-power
commercial LED device (LCW CQAR.EC) from OSRAM
Opto Semiconductors Company [9]. Also the figure shows
the spectral reflectance measured by the authors in [10] for
different materials in the visible band.
The multi-bounce power delay profile (PDP) in indoor
channel is given as [10]
h(t) =

N

LED 

h (k) (t; n )

(3)

n=1 k=0

where NLED is the number of LED sources and h (k) is the


PDP due to the kth reflection. If the LOS signal is lost due
to shadowing, then the received signal is the sum of reflections only. In [10], it is shown that in NLOS VLC, the first
reflection (k = 1) has clearly the largest contribution to the
received signal which contains about 80 % of the total power
as compared to subsequent order reflections. In our analysis of the impulse response, we have noticed that including

Photon Netw Commun

where 1 () is the spectral reflectance of the reflector. In this


paper, we assume the room is empty so the signal is reflected
by walls only.

nth LED

d1

2.3 System model

Blocked LOS

The received optical signal at the PD input is given by

1
2

LOS blocker

Pr = H (0)Pt

Shadow area

d2

where H (0) is the channel DC gain and Pt is the transmitted


power. The PD photocurrent is expressed as

FOV

(7)

Y (t) = R X (t) h(t) + N (t)

(8)

Wall Reflected Ray

where X (t) is the transmitted optical pulse, R is the detector


responsivity, and N (t) is the noise. The desired signal power
is given as

PD

Fig. 2 Geometry of the single reflection of the nth LED

the contribution of higher-order reflections (k > 1) to the


received signal requires complex computation. Therefore, we
primarily focus most of our attention to single reflection as
shown in Fig. 2, which is used to be the second most important optical path after the LOS. The channel response after
the first reflection to the nth LED is given as
h

(1)


2
(t; n ) =
FOV
s


d1 + d2
dAref
t
c


L 1 L 2 n(1) rec


Prs =
0

SNR =

The integration in (4) is performed over the whole surfaces


of the reflectors. Aref is the reflector area, which is the wall in
Fig. 2. APD is the photodetector area, c is the speed of light,
and FOV is the field of view of the detector. The photodiode
in Fig. 2 detects light if the incidence angle 2 FOV. The
rectangular function in (4) is given as

NLED

h i (t) X (t)dt

(10)

i=1

(6)

(11)

2 is the shot noise variance, and 2


where shot
therm is the thermal noise variance. The shot noise variance is given by
2
shot
= 2q R(Prs + Prisi )B + 2q Ibg I2 B

(12)

where q is the electron charge, B is the equivalent noise


bandwidth, Ibg is the background current, and I2 is the noise
bandwidth factor. The thermal noise is given as [7]
2
therm
=


n 1 ()d

(R Prs )2
2 + 2
2
shot
therm + (R Prisi )

(5)

The power of single reflection is given by


n(1) =

(9)

The quality of the VLC communication system is determined


by SNR with interference. The SNR is given by

Aref (m + 1) cosm 1 cos 1


,
2 d12
A cos 2 cos 2
L 2 = PD
d22

1, |x| 1
0, |x| > 1

h i (t) X (t)dt

2.4 SNR with intersymbol interference

L1 =

and the intersymbol interference (ISI) power contribution is


given by

(4)

NLED
i=1

Prisi =

where

rec(x) =

8 kTk
16 2 kTk 2 2
APD I2 B 2 +
APD I3 B 3 (13)
G
gm

where k is the Boltzmanns constant, Tk is the absolute temperature, G is the open-loop voltage gain, is the fixed
capacitance of PD per unit area, is the FET channel noise

123

Photon Netw Commun


Table 1 Parameters used in simulation
Parameter

Value

600

Transmitted power

0.7 W

Semi-angle at half power

60

Center luminous intensity

57 cd

Number of LED lamps

Illuminance(lx)

Transmitter

Number of LEDs/Lamp

49

Multi-transmitter position

(1, 1, 3); (1,4, 3); (4, 1, 3); (4, 4, 3)

500
400
300

5
4
3

Receiver
PD area

1 cm2

FOV

60

y (m)

Optical filter gain

Refractive index of a lens

1.5

Responsivity

0.54 A/W

Time resolution

0.2 ns

Data rate

40 Mbps (OOK NRZ)

M
=
2

40 Mb/s

Background current

5100 A

I2

0.562

Absolute temperature

298 K

Open-loop voltage gain

10

Fixed capacitance

112 pF/cm2

FET channel noise factor

1.5

FET transconductance

30 mS

I3

0.0868

x(m)

Fig. 3 Illuminance distribution on the room floor, Min = 300 lx;


Max = 550 lx; Avg. = 480 lx

Noise
Equivalent noise bandwidth

i=1


Prs,i ti2 + Nj=1 Prisi,i ti2
Prtotal

Using (14), the maximum data rate that can be transmitted


through the channel without equalizing is given by [3]
Rb

1
bps.
10Drms

(15)

3 Simulation results and discussion


factor, gm is the FET transconductance, and I3 is a constant.
The first term in (13) represents the feedback resistor noise,
while the second term represents the FET channel noise. The
values of the noise parameters are taken from [8] and listed
in Table 1.
2.5 Channel delay spread and maximum transmitted
data rate
The rms delay spread (Drms) is a critical criterion for system
performance. It determines the upper bound of data transmission rate. The Drms is calculated using the mean excess
delay () and is given by
Drms =


2 ()2

(14)

where
M
=

i=1

123

Prs,i ti +

N

Prtotal

j=1

Prisi,i ti

In this section, we evaluate the system performance when


LOS signal is absent. We consider an empty room with
dimensions 5 m 5 m 3 m. Four LED lamps are used.
Each lamp consists of 7 7 LED chips. The LED chip has a
PSD as shown in Fig. 1. The walls are assumed to be plastered
with spectral reflectance also shown in Fig. 1. The transmitter
parameters used in simulation are for the commercial product
LED (LCW CQAR.EC) and listed in Table 1.
3.1 Illuminance on room floor distribution
First, we investigate the horizontal illuminance at the room
floor. The sufficient illuminance as defined by ISO is between
300 and 1500 lx. In Fig. 3, we show the illuminance at the
room floor assuming no additional light reflected from surrounding walls. We notice that the illuminance is high at the
center and decreases at the edges. On average, the illuminance on the floor is 482 lx. The minimum at the edge is
300 lx and the maximum at the center is 550 lx. Therefore,
the LED lamps with the parameters used in this simulation
can guarantee sufficient illuminance.

Photon Netw Commun


44-85

1.7-2

120-170

85-120

10-19.2

y (m)

y (m)

5-10

2-5

x (m)

Fig. 4 Data rate distribution in Mbps at PD level: Min = 44 Mbps,


Max = 170 Mbps, Avg. = 84 Mbps

Fig. 5 Received power distribution in W at PD level (height =


0.85 m), Min = 1.7 W; Max = 19.2 W; Avg. = 7.2 W

12.7-14

3.2 Maximum transmitted data rate

14-18

18-28.5

y (m)

In Fig. 4, we calculated the maximum data rate distribution


that can be transmitted through the channel without need for
equalizers using (15). When the PD is located at equal distances from the transmitters, the Drms effect is lowest and
then the data rate is the highest. The maximum data rate that
can be achieved is at the middle of the room where all transmitters apart with equal distance. This is equal to 170 Mbps
with 0.6 ns Drms. The minimum data rate is 44 Mbps with
2.26 ns Drms. Therefore, for the rest of this paper, the data
rate is simulated at 40 Mbps.

x (m)

3.3 Received power distribution


The distribution of the received power with a PD oriented
toward ceil is shown in Fig. 5 where four regions are defined.
The PD is located on a table at height 0.85 m. Due to the
symmetry of the room and also the transmitters location, the
power distribution appears symmetric. High received power
is achieved at points located between two walls where more
power can be collected. The maximum received power is
19.2 W. The amount of received power decreases as the PD
location becomes far from the walls and the transmitters. The
worst case is when the PD is located far from the transmitter
in the middle of the room. The minimum received power by
a single detector is 1.7 W. This small power still can be
detected by a PD having sensitivity around 30 dBm. The
average of the received power over all points is 7.2 W. It is
also worthy to note that power distribution through the room
surface in our NLOS case is largely different from what is
used to be studied when LOS exists. For more details about

x (m)

Fig. 6 SNR in dB distribution at PD level: Min = 12.7 dB, Max =


28.5 dB, Avg. = 19 dB

power distribution in LOS system, the reader is referred to


[3,8].
3.4 SNR and BER distribution
Using the signal-to-noise ratio model in (11), we illustrate the
SNR distribution in the room at all possible PD locations as
shown in Fig. 6. Three regions are defined in Fig. 6. Since the
walls operate now as secondary light sources, the minimum
obtained SNR is 12.7 dB at the middle of the room (farthest
location to the walls). The highest SNR is observed close

123

Photon Netw Commun


<10-6

10-6 - 10-4

10-4 - 1.210-3

Tx2

Tx3

Tx1

Tx4

y (m)

LOS
blocking

x (m)

Fig. 7 BER distribution at PD level: Min = 1.2 103 , Max < 106

PD
Conf. #1

to the walls (secondary light sources) and specifically in the


corners where two walls reflections dominate the received
signal. In effect, the maximum SNR 28.5 dB is observed at
the corners. On average, 19 dB SNR is achieved through all
locations in the room.
In Fig. 7, we show the BER distribution for the system
obtained through Monte Carlo simulation of the NLOS system. The BER results are consistent with those of SNR in
Fig. 6 confirming that the quality of the communication may
very quickly change when the receiver changes its location
in the room for few tens of centimeters or less. A BER less
than 104 is observed almost everywhere in the room; 106
is mainly in the corners, and 103 in the center. Recall that
these regions resulted from the location we have fixed for the
LED lamps. The regions are very sensitive to the placement
plan of the LEDs. When observing the current general trend
of deploying LED lighting technology, it seems to develop
toward more distributed architectures through ceil surface,
rather than centralized. This gives more uniform illumination in the space and reduces energy consumption. Note this
is in the benefit of better uniformity of SNR and BER through
the room surface.

4 Multi-input multi-output (MIMO) system


In previous section, we investigated the performance of VLC
system with NLOS. Four LED lamps are used and only a
receiver with single LED. This system is called multi-input
single-output (MISO) system. In this section, we explore the
performance of VLC system when multi-PDs are used in the
receiver.

123

Conf. #2

Conf. #3

Fig. 8 A simulation for a room shows PD orientation

We define three operation configurations for the PDs in the


receiver. In configuration 1, single PD is used as discussed
in previous section. In configuration 2, four PDs oriented to
the room ceil are used, while in configuration 3 four PDs
are oriented to each wall in the room. The receiver device is
located on a table at 0.85 m height. The room configuration
and receiver position is shown in Fig. 8.
In configuration 1, the PD is placed at the center of the
receiver. In configurations 2 and 3, the distance between the
PDs and the receivers center is 10 cm. Selection combining
technique is used where only one PDs signal is selected at
a time that has the best SNR [11]. The system performance
is investigated at two receiver positions as shown in Table 2.
The first position corresponds to locations having minimum
SNR for configuration 1. The second position corresponds to
location having maximum SNR for configuration 1. For these
two positions, we investigate how MIMO system improves
the performance. The system performance for the three configurations in terms of SNR, BER, Drms, and data rate are
calculated and listed in Table 2.
The results show that when configuration 1 shows low
performance at location (2.3, 2.5, 0.85), especially in terms
of BER, the MIMO system in configuration 2 was able to
improve the system performance in terms of SNR, BER,
Drms, and data rate. The BER is improved by one order
of magnitude, while the data rate is improved by 9 Mbps. For
configuration 3, the performance was better in terms of SNR
and shows very low BER. However, the Drms was higher
which results in lower transmitted data rate.

Photon Netw Commun


Table 2 Simulation results for different PD positions
Position

Worst SNR
(2.3, 2.5, 0.85)

Best SNR
(0.2, 0.2, 0.85)

Drms (ns)

0.73

1.48

SNR (dB)

12.7

28.5

BER

1.2 103

<106

Data rate (Mbps)

135

67

Acknowledgments This research is supported by the KACST Technology Innovation Center in Radio Frequency and Photonics for the
e-Society (RFTONICS), King Saud University.

MISO Mode#1

MIMO Mode#2
Drms (ns)

0.64

1.65

SNR (dB)

14.5

28.1

BER

104

<106

Data rate (Mbps)

154

60

MIMO Mode#3
Drms (ns)

2.0

3.2

SNR (dB)

18.9

31

BER

<106

<106

Data rate (Mbps)

50

31

When configuration 1 shows high performance at location


(0.2, 0.2, 0.85), the MIMO system in configuration 2 shows
close performance to configuration 1. The performance of
configuration 1 was better because its location in the center of the receiver was better than the location of the PDs
in configuration 2. This enables configuration 1 to collect
more power and achieves a little higher SNR. Configuration 3 at this location shows higher SNR, but it increases the
Drms which results in lower data rate. Therefore, under NonLOS condition, single PD can work well in some locations
only. Therefore, MIMO system with PDs oriented toward
ceil is the best solution to achieve better performance in any
receiver position in the room. MIMO system with PDs oriented toward walls can achieve higher SNR and lower BER
than configurations 1 and 2, but it undergoes higher Drms
and lower data rate.

5 Conclusion
In this paper, the performance of VLC system under NLOS
condition is studied. The simulation results showed that VLC
system with NLOS signal has acceptable performance. We
studied three configurations for PD orientation in the VLC
receiver. The result showed that in general, the MIMO system
with PDs oriented to ceil can improve the system performance better than using single PD. The MIMO system with
PDs oriented toward walls has high Drms witch reduces the
maximum transmitted data rate.

References
[1] Saadi, M., Wattisuttikulkij, L., Zhao, Y., Sangwongngam, P.: Visible light communication: opportunities, challenges and channel
models. Int. J. Electron. Inform. 2(1), 111 (2013)
[2] OBrien, D., et al.: Visible light communications: emerging technologies and applications. In: Kraemer, M., Katz, M.D. (eds.)
Short-Range Wireless Communications: Emerging Technologies
and Applications. Wiley, Chichester (2009)
[3] Ghassemlooy, Z., Popoola, W., Rajbhandari, S.: Visible light
communications. In: Proceedings of Optical Wireless Communications: System and Channel Modelling with MATLAB (2013)
[4] Jovicic, A., Li, J., Richardson, T.: Visible light communication:
opportunities, challenges and the path to market. IEEE Commun.
Mag. 51(12), 2632 (2013)
[5] OBrien, D., Zeng, L., Le-Minh, H., Faulkner, G., Walewski, J.,
Randel, S.: Visible light communications: challenges and possibilities. In: Proceedings of the 19th International Symposium on
Personal, Indoor and Mobile Radio Communications (2008)
[6] Zongyu, D., Cui K., Chen G., Xu, Z.: Non-line-of-sight link performance study for indoor visible light communication systems.
In: SPIE Optical Engineering+ Applications. International Society for Optics and Photonics (2010)
[7] Barry, J.R.: Wireless Infrared Communications. Kluwer Academic, Dordrecht (1994)
[8] Komine, T., Nakagawa, M.: Fundamental analysis for visible-light
communication system using LED lights. IEEE Trans. Consum.
Electron. 50(1), 100107 (2004)
[9] Lcw, CQAR.EC, OSRAM opto semiconductors. http://www.osr
am-os.com/osram_os/en/products/product-catalog/leds-for-gen
eral-lighting/oslon-square/lcw-cqar.ec/index.jsp
(2015).
Accessed 18 Jan 2015
[10] Lee, K., Park, H., Barry, J.: Indoor channel characteristics for
visible light communications. IEEE Commun. Lett. 15(2), 217
219 (2011)
[11] Tuyen, L., Bao, V.: Comparison of diversity combining techniques
for MIMO systems. In: Proceedings of the 17th Asia-Pacific Conference on Communications (2011)

Maged A. Esmail received his B.E. degree


in electronic engineering from Ibb University in 2006 and M.Sc. degree (with first
class honors) in electrical engineering from
KSU University in 2011 where he is currently following his Ph.D. studies. From 2009
to 2013, he joined Prince Sultan Advanced
Research Technologies Institute (PSATRI) as
a researcher. From 2014, he joined KacstTechnology Innovation Center RTONICS as
an assistant researcher. His research interests include fiber-optic communications, PON and long-reach PON, network management and
protection, sensor networks, free space and visible light communications.

123

Photon Netw Commun


Habib A. Fathallah (S96, M01) received
the B.S.E.E. degree (with Honors) from the
National Engineering School of Tunis, 1994
and the M.A. and Ph.D. degrees in electrical and computer engineering from Laval
University, Qc, Canada, in 1997 and 2001,
respectively. He initiated the use of Bragg
gratings technology for all-optical/all-fiber
coding/decoding in Optical CDMA systems.
He was the founder of Access Photonic Net-

123

works (20012006) and adjunct professor with the Electrical and Computer Engineering Department of Laval University (Quebec, Canada).
He is currently with Electrical Engineering Department, College of
Engineering of the King Saud University (Riyadh, KSA), co-founder
and Photonics Thrust leader of Kacst-Technology Innovation Center
RTONICS. His research interests include coherent optical communications systems, mode and multicore space division multiplexing
(SDM), Free Space Optical and Visible light communications; Long
reach PONs, hybrid fiber wireless (FiWi) systems, and OCDMA.

Das könnte Ihnen auch gefallen