Open Access

A Novel Hybrid Dimming Control Scheme for

Visible Light Communications
Volume 9, Number 6, December 2017

Yang Yang, Student Member, IEEE

Zhimin Zeng
Julian Cheng, Senior Member, IEEE
Caili Guo, Senior Member, IEEE

DOI: 10.1109/JPHOT.2017.2771223
1943-0655 © 2017 IEEE
IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

A Novel Hybrid Dimming Control Scheme

for Visible Light Communications
Yang Yang ,1 Student Member, IEEE, Zhimin Zeng,2
Julian Cheng ,3 Senior Member, IEEE,
and Caili Guo ,1 Senior Member, IEEE

1 Beijing Laboratory of Advanced Information Networks, School of Information and

Communication Engineering, Beijing University of Posts and Telecommunications, Beijing
100876, China
2 Beijing Key Laboratory of Network System Architecture and Convergence, School of
Information and Communication Engineering, Beijing University of Posts and
Telecommunications, Beijing 100876, China
3 University of British Columbia, Kelowna, BC V6T 1Z4, Canada

DOI:10.1109/JPHOT.2017.2771223
1943-0655  C 2017 IEEE. Translations and content mining are permitted for academic research only.

Personal use is also permitted, but republication/redistribution requires IEEE permission.

See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Manuscript received October 13, 2017; accepted November 1, 2017. Date of publication November 8,
2017; date of current version December 8, 2017. This work was supported in part by the 111 project
(NO.B17007), in part by BUPT Excellent Ph.D. Students Foundation under Grant CX2016313, in part
by Huawei Innovation Research Program, in part by the National Science Foundation of China under
Grant 61271177, and in part by China Scholarship Council under Grant 201606470015. Corresponding
author: Yang Yang (e-mail: young0607@bupt.edu.cn).

Abstract: A dimming control scheme that combines analog dimming and spatial dimming is
proposed. The basic idea of the combined scheme is that both the direct current bias level of
each individual light-emitting diode (LED) and the number of glared LEDs are simultaneously
adjusted to achieve a required dimming level. Besides, we design a space constellation to
transmit additional information in spatial domain without compromising uniform illumination
constraint, and this essentially improves the performance of the system. A maximum-a-
posteriori probability estimator is also designed to estimate the received time domain and
spatial domain signals. Since two degree of freedoms are simultaneously exploited for
dimming control, the proposed scheme outperforms conventional dimming schemes, which
only focus on signal domain design, and this is verified by simulation results. To the best of
authors’ knowledge, this is the first design that effectively utilizes multidomain resources to
achieve dimming control for visible light communications.

Index Terms: Bit error rate, dimming control, orthogonal frequency division multiplexing,
visible light communication.

1. Introduction
Visible light communication (VLC) has gained increasing attentions in the past ten years. As an
access technology that combines illumination and communication, VLC needs to complete its illumi-
nation task first, and then tries to achieve better communication performance under the constraint
of illumination. This is also one of the most outstanding differences of VLC when compared to
conventional radio-frequency based communications [1], [2].
As an important illumination task, dimming control has been extensively studied. These studies
include analogue dimming schemes [3], [4], which change the optical power by varying the direct
current (DC) of the signal, and digital dimming schemes [5], which achieve required optical power
by controlling the duty cycle of signals. More recently, a new type of dimming method called spatial

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

dimming was proposed in [6], and the basic idea of spatial dimming is that the illumination levels can
be represented by the number of light-emitting diodes (LEDs) in an LED lamp. Note that since the
power of a single LED is limited, multiple LEDs are often used to achieve sufficient illumination level.
Even though these three types of dimming schemes have been carefully designed, they have
their own drawbacks. Analogue dimming has relatively severe chromatic distortion problem [5], as
DC bias is varied for different dimming levels. Besides, the amplitude of the signal in low dimming
level will be significantly confined by the dynamic range of LEDs, which may lead to undesired
communication performance. As for spatial dimming schemes, since the dimming level for each
LEDs is fixed, a large number of LEDs are required to achieve precise dimming control. Since
the number of LEDs in an LED lamp is pre-determined, this limitation may significantly affect the
applications of spatial dimming.
To address the shortcomings of the aforementioned dimming control schemes, a novel hybrid
dimming (HD) scheme is proposed in this work for a typical indoor scenario. Note that the HD
scheme applies to a variety of modulations due to the orthogonality between signal domain and
spatial domain. We adopt direct current biased orthogonal frequency division multiplexing (DCO-
OFDM) in this paper to show the superiority of HD-OFDM. The basic idea of HD-OFDM is that both
the DC biasing level and the number of glared light-emitting diodes can be adjusted to achieve a
certain dimming level. On the one hand, the proposed HD schemes may partly mitigate chromaticity
shift and clipping noise compared to AD schemes, as part of illumination adjustments are achieved
in the spatial domain. On the other hand, HD schemes can achieve precise dimming control
without any constraint on the number of LEDs due to the DC biasing adjustment. Furthermore,
we design a space constellation for the dimming system which can transmit additional information
by activating a certain subset of LEDs without compromising uniform illumination requirement.
Moreover, a maximum-a-posteriori probability (MAP) based receiver is designed for the proposed
system. Simulation results will verify that the hybrid design can significantly enhance the bit-error
rate (BER) performance of the system, thus achieving higher energy efficiency.
It should be noted that there are some previous studies focusing on combining optical spatial
modulation and optical OFDM schemes (OSM-OFDM) [7]–[9]. For example, in [7] the sign of OFDM
signals are encoded in LED indexes, thus avoiding spectrum loss in time domain. Similarly, in [8],
the DC bias of OFDM signals is avoided as positive signals and negative signals (after inverse) are
transmitted by two LEDs, separately. In [9], the performance of optical OFDM schemes with both
spatial modulation and spatial multiplexing are compared using three different front-ends. However,
all these studies only focused on the communication aspect and they have some limitations in terms
of dimming control. For instance, in [7] only two out of four LEDs can be activated at one time. This
means the maximum normalized optical power for the scheme is only 50%. The contributions of
this paper are listed as follows.
1) In terms of the aforementioned dimmable optical OFDM schemes, this is the first design that
effectively utilizes multi-domain resources (i.e. signal domain and spatial domain resources)
to achieve dimming control in VLC systems.
2) In terms of the previous studies on OSM-OFDM systems, this paper proposes a generalized
OSM-OFDM scheme, termed as HD-OFDM, which can guarantee uniform illumination and
achieve high communication performance in a wide dimming range.
3) A MAP detection algorithm is also proposed for the generalized OSM-OFDM system, which
can be effectively applied regardless of the variation of dimming levels.
The reminder of this paper is organized as follows. In Section 2, we introduce the system model
of the proposed HD scheme. Spatial domain design and the MAP estimator are detailed in Sec-
tion 3. Simulation results are presented and discussed in Section 4. Finally, Section 5 draws some
important conclusions.

2. The Proposed HD-OFDM Scheme

As shown in Fig. 1(a), we consider a 4 × 4 MIMO system in a 4 m × 4 m × 3 m room [7], [10].
The input bits are split into two parts according to Fig. 1(b). For the part conveyed to the signal

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

Fig. 1. (a) The considered geometric scenario. (b) Diagram of the proposed HD-OFDM scheme.

domain, they are modulated according to DCO-OFDM. First, Hermitian symmetry is imposed into
the quadrature amplitude modulation signals as

X (k) = X ∗ (N − k),
(1)
X (0) = X ( N2 ) = 0

k = 1, 2, . . . N − 1, where X (k) is the frequency-domain OFDM signals, N is the number of subcar-

riers. After the inverse fast fourier transformation (IFFT) operation, the time-domain OFDM signals
−1
x(n) = √1N Nk=1 X (k) exp{ j2πnk
N
} are real due to the Hermitian symmetry. Following the cyclic prefix
insertion, the transmitted useful signal can further turn into positive signals after adding DC biasing
as
s(n) = ax(n) + B (2)
where a is a scaling factor, and B is the biasing level. Since signals out of the dynamic range of
LEDs will be clipped, the transmitted signal can be expressed as
⎧
⎨I h,
⎪ s(n) ≥ I h
z(n) = s(n), I l ≤ s(n) < I h (3)
⎪
⎩
I l, I l > s(n)
where z(n) is the transmitted signal, I h and I l are the maximum and minimum allowed currents of
an LED. Then a matrix T = [t i ]N t ×1 , which is determined by the input bits of the spatial domain,
is imposed to modulate the spatial domain signal, where N t is the number of transmitters, which
equals 4 in this paper. Let t i ∈ {0, 1}, where t i = 1 means the i -th LED is active; t i = 0 means the
i -th LED is inactive. At the receiver side, the detected matrix Y can be denoted as
Y = HTz + W
⎡ ⎤⎡ ⎤ ⎡ ⎤
h 11 . . . h 1N t t1 w1
⎢ . ⎥⎢ . ⎥ ⎢ . ⎥ (4)
=⎢⎣ ..
.
.. .. ⎥
. ⎢ ⎥ ⎢
⎦ ⎣ .. ⎦ z + ⎣ ..
⎥
⎦
hNr1 . . . hNrNt tN t wNr
where H denotes the channel gain matrix, W consists of the additive white Gaussian noise (AWGN).
The channel gain of a line of sight (LOS) optical propagation link can be represented as follows:

(m+1)A
g(ψ) cosm (φ) cos (ψ) , 0 < ψ < c
h los = 2πd 2 (5)
0, ψ > c

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

where A is the detector area; d is the distance between the transmitter and the receiver. Furthermore,
m is the Lambert order defined as m = − ln 2/ ln(cos 1/2 ), where 1/2 is the transmitter semiangle
(at half power); ψ is the angle of incidence; φ is the angle of irradiance; c is the receiver field of
vision (FOV) semiangle and g(ψ) denotes the gain of optical concentrator and it can be denoted
as
 n2
r
, 0 ≤ ψ ≤ c ,
g(ψ) = sin c
2
(6)
0, ψ ≥ c
where n r is the refractive index. Besides LOS links, there are also non-LOS links which can be
bounced from wall, ceil and floor before they reach the receiver. Since the non-LOS links may also
affect the system performance, they are taken into consideration in this paper. The diffuse channel
loss can be expressed as [11], [12]
A ρ1
h diff = (7)
A room 1− < ρ >
where A room is the surface of the room, ρ1 is the reflectivity of the region initially illuminated by
LEDs, <ρ> is the average reflectivity and it can be denoted as
1 
<ρ> = A i ρi (8)
A room i

where A i are the areas of different surface such as wall, ceil or floor, and ρi are their corresponding
reflectivity. Since the root mean delay between LOS and non-LOS links are in nano seconds [1],
[11], [13] while typical bandwidth of off-the-shelf LEDs are few megahertz [1], flat channel are often
considered for indoor VLC [4], [5], [10]. Furthermore, for O-OFDM systems, it has longer symbol
period compared to single-carrier modulation such as pulse amplitude modulation, and cyclic prefix
(CP) is further used to mitigate inter-symbol interference (ISI) and inter-carrier interference [13],
[14]. Note that ISI resistance is one of the major advantages of OFDM system. Thus a flat channel
is considered and we have
h = h los + h diff (9)
in this paper. At the receiver side, a MAP estimator is designed to extract the information conveyed
by the signal domain and spatial domain, respectively, which will be specified in Section 3.

3. Transceiver Design of HD-OFDM

Since the biasing level and the number of glared LEDs N a are simultaneously adjusted, we have
N a ≤ N t in HD-OFDM. Utilizing this fact, information can also be transmitted by selecting the glared
LEDs. The achieved normalized optical power can be represented as [3]
N aB
η= . (10)
(I l + I h )N t
Note that in (10), both B and N a are variables and they can be adjusted according to the target
dimming level η. Since N a ≤ N t , r = log2 (( NN at )) bits can be conveyed using the generalized spatial
modulation, where · is the floor function. Note that there are ( NN at ) LED patterns that can transmit
spatial domain signals while 2r patterns are sufficient to take full use of spatial domain resources.
This means there are extra ( NN at ) − 2r patterns that should be removed from the potential patterns.
Therefore, the matrix T should be carefully designed to select proper LED patterns.

3.1 The Design of Spatial Matrix T

Generally, uniform-distributed illuminance and signals-to-noise ratio (SNR) are preferred in indoor
lighting [15]. These two constraints should not be compromised when implementing dimmable VLC

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

TABLE 1
HD-OFDM spatial mapping table, where numbers in (.) indicates the active antenna indices

Input bits Na = 1 Na = 2 Na = 3

00 (1) (1, 2) (1, 2, 3)

10 (2) (2, 3) (1, 3, 4)
01 (3) (3, 4) (1, 2, 4)
11 (4) (1, 4) (2, 3, 4)

schemes. To achieve this target, each LED should be activated with equal probability. In this way,
the proposed hybrid dimming scheme can have the same illuminance performance as conventional
dimming schemes. Thus, two cases need to be considered.
Case 1: log2 ( NN at ) = r . In this case, all the ( NN at ) combinations are used for transmissions. Therefore,
it is intuitive that all the LEDs have the same probability to be activated. For the scenario consid-
ered (N t = 4), Case 1 includes the situations when N a = 1, 3, where log2 ( NN at ) = r = 2. As a result,
mapping the LEDs according to the input bits with one-one correspondence can satisfy the uniform
demand.
Case 2: ( NN at ) > 2r . In this case, only 2r patterns should be selected as potential active patterns.
A basic principle is that, the activating probability for each LED should be the same, to achieve
uniform illumination and communication coverage. For the case when N t = 4, the situation that
N a = 2 belongs to Case 2. One of the possible mapping is listed in Table 1.
When N a = 4, all LEDs are constantly activated. As a result, no information can be transmitted
in spatial domain. The proposed scheme actually degenerates to conventional analogue dimming
scheme in this case. Therefore, it can be concluded that the analogue dimming scheme is a special
case of the proposed hybrid dimming scheme. It should be noted that since the bandwidth of typical
LEDs is in megahertz while flicker faster than 200 hertz can avoid any harmful effects [1], the
transformation between active and inactive states in HD-OFDM will not cause any side effects.

3.2 MAP Estimator for HD-OFDM

A straightforward solution to decode HD-OFDM signals is zero-forcing (ZF) equalizer [7], which
has been broadly used in current studies such as [8], [9]. Despite the simplicity, ZF equalizer can
significantly amplify the noise part due to the inverse of correlated channel matrix. To mitigate this
problem, a MAP estimator is designed. For a given matrix T, a conditional MAP estimator can be
expressed as

z̄ = arg max p (z|Y)

z
(11)
= arg max p (Y|z)p (z).
z

Clipped by the dynamic range of LEDs, the probability density function (PDF) of z can be denoted
as
⎧ I h −B
⎪ Q ( aσ ), z = Ih
⎪
⎨  
2
p (z) = √ 1 exp − (z−B )
, Il < z < Ih (12)
⎪
⎪ 2πaσ 2a 2 σ2
⎩ −B
1 − Q ( I laσ ), z = Il

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

where Q (·) is the Gaussian Q -function, and σ is the standard deviation of the useful signal x(n).
Since the AWGN channel is considered, the conditional PDF is
⎧  2 ⎫
⎪
⎪    ⎪
Y− h j z  ⎪
Na
⎪
⎪ ⎪
⎨  gi  ⎪⎬
1 i =1
p (Y|z) =  N r /2 exp ⎪− (13)
2πσw2 ⎪ 2σw2 ⎪
⎪
⎪
⎪ ⎪
⎪
⎩ ⎭

where N r is the number of antennas at  the receiver, σw is the standard deviation of AWGN, and
j j j

·
denotes the Frobenius norm, G j = g 1 , g 2 . . . , g N a , j = 1, 2 . . . , 2r stands for the j-th index
combination of the active LEDs. Simply by plugging z = I l and z = I h into p (Y|z)p (z), we can obtain
the MAP metric in these two cases. Then, we focus on obtaining the maximum MAP estimation
metric under the constraint I l < z < I h . After some simplifications, the corresponding MAP metric
for I l < z < I h is
 2
 
Y −  h j z 
Na
 gi 
(z − B )2
i =1
z̃ j = arg min +
z 2σw2 2a 2 σ2
⎡ T N ⎤
Na  a
σ 2
= arg min ⎣ hg j hg j + 2 w 2 ⎦ z 2
z i i a σ
i =1 i =1
 

Na
2B σ2 B 2 σw2
− 2YT
hg j + 2 2w z + YT Y + (14)
i a σ a 2 σ2
i =1

where (·)T is the transpose operation. Since

⎡ T N ⎤

Na  a
σ 2
⎣ hg j hg j + 2 2 ⎦ > 0
w
i i a σ
i =1 i =1

and the MAP estimation metric is a quadratic equation of z, we have

N T
a
a σ
2 2
hg Y + B σw2
j
i
i =1
z̃ j = N T N . (15)
 a  a
a 2 σ2 hg j hg j + σw2
i i
i =1 i =1

Plugging z̃ j in p (Y|z)p (z), we have the overall MAP metric as

⎧ ⎧  2 ⎫
⎪
⎪ ⎪
⎨ Y−  h j I h  ⎪
 Na 
⎪
⎪   ⎬
⎪
⎪ M A P 1
= 1
exp − i =1
gi
−B
Q ( I haσ ), z = Ih
⎪
⎪ j
(2πσw2 )
N r /2
⎪ 2σw 2
⎪
⎪
⎪ ⎩ ⎭
⎪
⎪ ⎧
⎪
⎪  2 ⎫
⎪
⎨ ⎪
⎨ Y−  h j z̃ j  ⎪
 Na 
⎬
(z̃ j −B )2  gi 
p (z)p (Y|z) = M A P j = (2π)(N r +1)/2 (aσ)1/2 σN r exp − 2a 2 σ2 −
2 1 i =1
, Il < z < Ih (16)
⎪
⎪ w ⎪
⎩
2σw2 ⎪
⎭
⎪
⎪
⎪
⎪ ⎧  2 ⎫
⎪
⎪ ⎪ Y− h j I l  ⎪
 
Na 
⎪
⎪   ⎨ ⎬
⎪
⎪ I l −B
 gi 
⎪
⎪ M A P 3
= 1
1 − Q ( ) exp − i =1
, z = I l.
⎪
⎩
j
(2πσw2 )
N r /2 aσ ⎪
⎩
2σw 2
⎪
⎭

The detection algorithm is depicted in Algorithm 1. After these operations, z̄ is exported to FFT
block to recover frequency-domain OFDM signals while G j̃ is demodulated by spatial demodulation
block according to space constellation.

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

Algorithm 1: The MAP estimator for HD-OFDM.

1: Input: Y, N a , r
2: for j = 1 : 2r do
3: for i = 1 : 3 do
4: Obtain M A P ji according to (16);
5: end for
6: end for
7: Obtain the maximum MAP metric by (˜i , G j̃ ) = arg max M A P ji
i ,G j
8: if ˜i == 1 then
9: z̄ = I h
10: else {˜i == 2}
11: Set z̄ = z̃ j according to (15)
12: else {˜i == 3}
13: z̄ = I l
14: end if
15: Output: z̄, G j̃ .

4. Simulation Results
In this section, we compare the BER performance of HD-OFDM with analogue dimming based DCO-
OFDM scheme for varied configurations. More specifically, the proposed HD-OFDM scheme with
both ZF equalizer (HD-ZF) and MAP estimator (HD-MAP) is investigated. The spectral efficiency of
2 M (N −2)
DCO-OFDM is ηD CO = log 2(N +N cp )
[16]. Conveying extra transmission in spatial domain, its data rate
is improved as
    
log2 M (N − 2) Nt
R H D −O F D M = B w + log2 (17)
2(N + N cp ) Na
where N cp is the length of the cyclic prefix, and B w is the bandwidth of LEDs. For the ana-
logue counterparts, repetition coding and spatial multiplexing based DCO-OFDM schemes with
ZF equalizers (RC-ZF and SMP-ZF) are also studied, and their data rates can be denoted as
2 M (N −2) 2 M (N −2)
R R C−Z F = B w log
2(N +N cp )
and R SM P −Z F = B w N t log
2(N +N cp )
, respectively. Note that both for RC and SMP
based schemes, all the LEDs are activated, and they adjust dimming levels by changing their DC
biasing levels. Since the same scenario is considered as that in [10], we mainly employ the same
system parameters as in [10] if not otherwise specified, which are shown in Table 2. Note that both
the transmitter and the receiver are in the center of the room with 1.75 meter vertical distance [10].
Similar to the definition in [17], we set the scaling factor a as
min(B − I l , I h − B )
a= ! (18)
σ 10A dB /10 − 1
where A dB is set to 13 dB in this paper. N a then can be optimized by simulations.
Setting fixed data rate as 11.6 Mbps for all the compared schemes, we first compare the BER
performance of DCO-OFDM and HD-OFDM schemes at low dimming levels. As shown in Fig. 2,
HD-OFDM outperforms DCO-OFDM due to the utilization of spatial resources. For instance, HD-
ZF has 10 dB gains over RC-ZF when BER equals 10−3 and η = 10%. Furthermore, due to the
avoidance of amplifying noise, HD-MAP can further improve the BER performance over 3 dB, thus
gaining extra energy efficiency. It is also shown that SMP based system has the worst performance
due to the limited energy of each independent link.
Under the same configuration and data rate constraint, we evaluate the BER performance of
the considered schemes at higher dimming levels. As shown in Fig. 3, HD-MAP system can still
obtain substantial performance gains when compared to other systems when η = 50%. However,

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

TABLE 2
System specifications

Name of Parameters Values

The maximum allowed current, I h 2A

The minimum allowed current, I l 0A
The bandwidth of the LEDs, B w 4M
The length of the cyclic prefix, N cp 4
Floor reflectivity 0.15
Ceiling reflectivity 0.8
Wall reflectivity 0.7
height of the transmitter 2.5 m
height of the receiver, h r 0.75 m
The electrical power of the signal, σ2 1W
Refractive index, n r 1.5
Semiangle at half power, 1/2 15 [deg.]
The detect area of the PDs, A 1 [cm2 ]
Receiver FOV semiangle, c 15 [deg.]
The distance between PDs 0.1 m
The distance between LEDs, d t 0.6 m

Fig. 2. The BER performance of HD-OFDM and DCO-OFDM for η = 10%, 30%.

the performance gain shrinks when compared to that with lower dimming levels. This is because,
when the dimming level is high enough, all the LEDs should be activated to satisfy the required
dimming level, which makes it impossible to utilize spatial domain resources for transmissions. This
explains why HD-OFDM schemes degenerate to conventional RC based DCO-OFDM schemes
when η = 70%, i.e., they have the same BER performance. For practical indoor lighting scenarios,
the employed dimming level is both related to the users’ preference and the lighting requirement.

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

Fig. 3. The BER performance of HD-OFDM and DCO-OFDM for η = 50%, 70%.

Fig. 4. The BER performance of HD-OFDM and DCO-OFDM for varied d t when η = 10%.

Thus, the proposed scheme can essentially improve the performance of conventional dimming
scheme in most of the cases if it is assumed that all the dimming levels have equal probability to
be adopted. More importantly, in many scenarios, the optical power is required to be low for the
purpose of energy saving [18] or eye safety insurance [19] (especially in uplink transmission [20]).
Therefore, it is meaningful to enhance the performance of the dimming scheme at low dimming
levels, which will be further studied next.
We evaluate the BER performance with varied d t in Fig. 4 by fixing E b /N 0 = 27 dB and data rate
as 15.3 Mbps. First of all, it should be noted that smaller d t indicates higher channel correlations and
higher channel gains. Besides, MAP estimator outperforms ZF in high channel correlation scenario
because it avoids inverting channel matrix and can utilize the correlated channels for the estimation.
With these observations, it is shown that the BER of HD-MAP, HD-ZF and SMP-ZF go down with the
increase of d t because all these schemes can benefit from the decrease of channel correlation. Two

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

Fig. 5. The BER performance of HD-OFDM and DCO-OFDM versus the height of the receiver when
η = 10%.

exceptions can be found. The first one is that when d t ≤ 0.55, HD-ZF have the same performance
as RC-ZF, and their BERs gradually increase for d t ≤ 0.55. This is because the channel correlation
is high for d t ≤ 0.55, thus HD-ZF degenerates to RC-ZF (i.e. all LEDs are active) to obtain better
performance. For RC-ZF, its performance degenerates with the decrease of channel gain. Thus
the performance of HD-ZF also degenerates in this region. The second exception is that when
d t ≥ 0.7, the BER of HD-MAP increases. This is because when employing the configuration in [10]
and d t ≥ 0.7, the system is aligned which means one transmitter and receiver are in each others’
field of vision (FOV) and channel correlation is eliminated [10]. Since the main performance gain of
the proposed MAP estimator is that it can better utilize the correlated channels for estimation, this
gain vanishes when no correlated elements are in the channel matrix. Furthermore, since the total
channel gain dramatically decreases when the system is aligned and no correlated channel exists,
the performance of HD-MAP degrades at d t = 0.7.
In Fig. 5, we evaluate the performance of the considered systems with varying transmission
distances when E b /N 0 = 27 dB and data rate equals 15.3 Mbps. The transmission distance is
adjusted by changing the height of the receiver. For lower h r , the channel correlation is higher due
to the longer distance. This can be intuitively explained as follows. For longer transmission distance,
the distance between LEDs and the distance between PDs are relatively shorter, leading to higher
channel correlation. Besides, for the considered configuration, the channel gain is decreased with
higher h r values. This phenomenon can be explained from (5), the channel gain of an optical link not
only depends on the transmission distance, but also depends on the angle of incidence ψ and angle
of irradiance φ. Since increasing h r makes ψ and φ larger, from (5), shorter transmission distance
may also result in lower optical gain. With these observations, we can see that the BER of HD-MAP
first goes down with the increase of h r due to the decreasing channel correlation. However, with
the increase of h r as well as ψ and φ, the system is aligned when h r ≥ 0.9. The BER of HD-MAP
goes up due to the decrease of total channel gain. As for HD-ZF, it degenerates to RC-ZF in low h r
regions due to the high correlated channels. Their BERs slowly go up as channel gain decreases.
When the correlation is low enough, HD-ZF can also utilize the spatial domain resource, thus its
BER decreases for larger h r values. From the illustrations of Figs. 4 and 5, we can conclude that
there is a trade-off between the decrease of the channel correlation and the reduction of channel
gains. The performance of HD-OFDM can be optimized when the distance between LEDs and the
distance between PDs and LEDs are carefully designed.

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

Fig. 6. The BER performance of HD-OFDM and DCO-OFDM versus the data rate when η = 10%.

In Fig. 6, we compare the performance of HD-OFDM and DCO-OFDM when the data rate is
increased for E b /N 0 = 27 dB. The data rate is improved by increasing the modulation order of the
quadrature amplitude modulation. As shown in Fig. 6, BER increases with the data rate. This is
because with the increasing of the modulation order, higher signal-to-noise ratio is required for a
certain BER. It can be observed that HD-MAP can significantly outperform the other schemes, and
HD-ZF also has lower BER when compared to the conventional schemes.

5. Conclusion
A novel dimming scheme that effectively combines analogue dimming and spatial dimming is pro-
posed. Utilizing the spatial domain resource, the proposed scheme can convey extra information
in spatial domain without affecting the uniform illumination constraint, leading to better BER perfor-
mance for a fixed spectral efficiency. Furthermore, a MAP estimator is specially designed, which
can further improve the performance of the proposed scheme. In conclusion, this paper proves the
feasibility and demonstrates the advantages of utilizing multi-domain resources for dimming con-
trol, indicating that this joint design is promising for future dimmable visible light communications.
Experimental verification is our future goal which is now being undertaken.

References
[1] P. H. Pathak, X. Feng, P. Hu, and P. Mohapatra, “Visible light communication, networking, and sensing: A survey,
potential and challenges,” IEEE Commun. Surveys Tut., vol. 17, no. 4, pp. 2047–2077, Fourth Quarter 2015.
[2] D. Wang, P. Ren, Q. Du, L. Sun, and Y. Wang, “Security provisioning for MISO vehicular relay networks via cooperative
jamming and signal superposition,” IEEE Trans. Veh. Technol., to be published.
[3] Y. Yang, Z. Zeng, J. Cheng, and C. Guo, “An enhanced DCO-OFDM scheme for dimming control in visible light
communication systems,” IEEE Photon. J., vol. 8, no. 3, Jun. 2016, Art. no. 7904813.
[4] Q. Wang, Z. Wang, and L. Dai, “Asymmetrical hybrid optical OFDM for visible light communications with dimming
control,” IEEE Photon. Technol. Lett., vol. 27, no. 9, pp. 974–977, May 2015.
[5] H. Elgala and T. D. C. Little, “Reverse polarity optical-OFDM RPO-OFDM: Dimming compatible OFDM for gigabit VLC
links,” Opt. Exp., vol. 21, no. 20, pp. 24288–24299, Oct. 2013.
[6] Y. Yang, Z. Zeng, J. Cheng, and C. Guo, “Spatial dimming scheme for optical OFDM based visible light communication,”
Opt. Exp., vol. 24, no. 26, pp. 30254–30263, Dec. 2016.
[7] A. Yesilkaya, E. Basar, F. Miramirkhani, E. Panayirci, M. Uysal, and H. Haas, “Optical MIMO-OFDM with generalized
LED index modulation,” IEEE Trans. Commun., vol. 65, no. 8, pp. 3429–3441, Aug. 2017.

Vol. 9, No. 6, December 2017 7908312

IEEE Photonics Journal Novel Hybrid Dimming Control Scheme

[8] Y. Li, D. Dobroslav, and H. Hass, “Non-DC-biased OFDM with optical spatial modulation,” in Proc. IEEE 24th Annu. Int.
Symp. Personal, Indoor, Mobile Radio Commun., 2013, pp. 486–490.
[9] C. He, T. Q. Wang, and J. Armstrong, “Performance comparison between spatial multiplexing and spatial modulation
in indoor MIMO visible light communication systems,” in Proc. IEEE Int. Conf. Commun., 2016, pp. 22–27.
[10] T. Fath and H. Haas, “Performance comparison of MIMO techniques for optical wireless communications in indoor
environments,” IEEE Trans. Commun., vol. 61, no. 2, pp. 733–742, Feb. 2013.
[11] L. Zeng, D. O’Brien, H. Le-Minh, K. Lee, D. Jung, and Y. Oh, “Improvement of date rate by using equalization in an
indoor visible light communication system,” in Proc. 4th IEEE Int. Conf. Circuits Syst. Commun., 2007, pp. 678–682.
[12] V. Jungnickel, V. Pohl, S. Nonnig, and C. von Helmolt, “A physical model of the wireless infrared communication
channel,” IEEE J. Sel. Areas Commun., vol. 20, no. 3, pp. 631–540, Apr. 2002.
[13] S. Dimitrov, S. Sinanovic, and H. Haas, “Clipping noise in OFDM based optical wireless communication systems,” IEEE
Trans. Commun., vol. 60, no. 4, pp. 1072–1081, Apr. 2012.
[14] S. Dimitrov and H. Haas, “Information rate of OFDM-based optical wireless communication systems with nonlinear
distortion,” J. Lightw. Technol., vol. 31, no. 6, pp. 918–929, Mar. 2013.
[15] Z. Wang et al., “Performance of a novel LED lamp arrangement to reduce SNR fluctuation for multi-user visible light
communication systems,” Opt. Exp., vol. 20, no. 4, pp. 4564–4573, Feb. 2012.
[16] D. Tsonev, S. Videv, and H. Haas, “Unlocking spectral efficiency in intensity modulation and direct detection systems,”
IEEE J. Sel. Areas Commun., vol. 33, no. 9, pp. 1758–1770, Sep. 2015.
[17] J. Armstrong and B. Schmidt, “Comparison of asymmetrically clipped optical OFDM and DC-biased optical OFDM in
AWGN,” IEEE Commun. Lett., vol. 12, no. 5, pp. 343–345, May 2008.
[18] I. Din and H. Kim, “Energy-effcient brightness control and data transmission for visible light communication,” IEEE
Photon. Technol. Lett., vol. 26, no. 8, pp. 781–784, Apr. 2014.
[19] M. B. Rahaim, A. M. Vegni, and T. D. C. Little, “A hybrid radio frequency and broadcast visible light communication
system,” in Proc. GLOBECOM Workshops, 2011, pp. 792–796.
[20] Y. Yang, Z. Zeng, J. Cheng, and C. Guo, “An amplify-and-forward based OFDM system for vlc uplink transmission,” in
Proc. IEEE Global Commun. Conf., to be published.

Vol. 9, No. 6, December 2017 7908312