Northumbria Research Link
Citation: Eltokhey, Mahmoud Wafik, Khalighi, Mohammad-Ali and Ghassemlooy, Zabih (2019) Multiple
Access Techniques for VLC in Large Space Indoor Scenarios: A Comparative Study. In: 2019 15th
International Conference on Telecommunications (ConTEL): Graz, Austria, 3-5 July 2019. IEEE,
Piscataway, NJ, pp. 1-6. ISBN 9781728120928, 9781728120911
Published by: IEEE
URL: https://doi.org/10.1109/contel.2019.8848536 <https://doi.org/10.1109/contel.2019.8848536>
This version was downloaded from Northumbria Research Link: http://nrl.northumbria.ac.uk/40948/
Northumbria University has developed Northumbria Research Link (NRL) to enable users to access
the University’s research output. Copyright © and moral rights for items on NRL are retained by the
individual author(s) and/or other copyright owners. Single copies of full items can be reproduced,
displayed or performed, and given to third parties in any format or medium for personal research or
study, educational, or not-for-profit purposes without prior permission or charge, provided the authors,
title and full bibliographic details are given, as well as a hyperlink and/or URL to the original metadata
page. The content must not be changed in any way. Full items must not be sold commercially in any
format or medium without formal permission of the copyright holder. The full policy is available online:
http://nrl.northumbria.ac.uk/policies.html
This document may differ from the final, published version of the research and has been made
available online in accordance with publisher policies. To read and/or cite from the published version
of the research, please visit the publisher’s website (a subscription may be required.)
Multiple Access Techniques for VLC in Large
Space Indoor Scenarios: A Comparative Study
1st Mahmoud Wafik Eltokhey, 2nd Mohammad-Ali Khalighi
3rd Zabih Ghassemlooy
Aix-Marseille University, CNRS,
Centrale Marseille, Institut Fresnel
Marseille, France
Mahmoud.Eltokhey@fresnel.fr
Ali.Khalighi@fresnel.fr
Optical Communications Research Group
Faculty of Engineering and Environment
Northumbria University
Newcastle upon Tyne NE1 8ST, UK
Z.Ghassemlooy@northumbria.ac.uk
Abstract—The growing demand for high speed indoor wireless
connectivity is among the driving forces for data transmission
based on visible-light communications (VLC). For relatively
large-space indoor scenarios, the development of appropriate
spectrally-efficient multiple-access (MA) techniques enables efficient handling of multiple users, in particular, in dealing with
the limited modulation bandwidth of the light-emitting diodes.
In this paper, we present a comparative study between different
MA techniques proposed in the recent literature for VLC
networks. The most appropriate schemes for large-scale network
deployments are further investigated in different scenarios to
contrast their performance in terms of the achievable throughput.
Index Terms—Visible light communications; multiple-access
techniques; indoor VLC; OFDMA; NOMA.
I. I NTRODUCTION
Visible light communications (VLC) has been receiving
increasing interest in the recent years, due to several promising
features including: using an unregulated spectrum, immunity
to radio-frequency (RF) interference, improved security at
the physical layer compared to the RF wireless technologies, and exploiting the energy-efficient light-emitting diode
(LED) lighting infrastructure [1]. Within this context, for
large-space indoor scenarios such as lecture halls, libraries,
convention centers, factories, airports, and shopping malls,
the design and deployment of the VLC technology faces a
number of challenges. In particular, handling a large number
of users involving a heavy wireless traffic, and managing user
mobility through multi-cell architectures with seamless handover, require efficient spectrum management and proactive
coordination between the access points (APs), i.e., the LED
luminaires, and the user terminals. There, an important consideration concerns the design of efficient multiple-access (MA)
techniques of reasonable complexity, in order to optimize the
network performance through the minimization of the intercell interference (ICI), as well as the inter-user interference
(IUI) within a cell.
Multi-user (MU) VLC system design has been the subject
of extensive research in the recent literature. Considering the
general context of VLC applications, a general overview of the
MU VLC techniques was presented in [2]. Also, a survey of
different MA techniques for VLC applications was provided
in [3] and [4]. Authors in [5] reviewed the non-orthogonal
multiple access (NOMA) approach for VLC. The work in [6]
studied suitable MA techniques for ultra-dense VLC networks
with a focus on NOMA and orthogonal multiple-access techniques. However, these works did not specially study the largespace scenarios. Our aim in this paper is to focus on the largespace indoor scenarios and to investigate the performance
of the most relevant MA techniques. In particular, we study
the effect of increased number of users on the maximum
achievable network data-rate.
The remainder of this paper is organized as follows. Focusing on downlink transmission, Section II presents different MA
techniques that could be used in VLC applications and discusses the suitability of each technique for use in large-space
VLC scenarios. Focusing on the most suitable techniques,
which are orthogonal frequency-multiplexing modulation MA
(OFDMA) and NOMA, as we will see, Section III presents the
mathematical formulation of signal transmission. Afterwards,
the performance of OFDMA and NOMA are contrasted in
Section IV for typical scenarios through numerical results.
Section V concludes the paper.
II. M ULTIPLE ACCESS T ECHNIQUES FOR VLC
We present here different MA techniques, namely, timedivision MA (TDMA), space-division MA (SDMA), optical
code-division MA (O-CDMA), OFDMA, and NOMA, while
discussing their suitability for a multi-cell VLC network with
the possibility of user mobility. Corresponding to each cell, an
LED luminaire serves as AP, handling the signal transmission
for users in its coverage area.
A. TDMA
By TDMA, the channel temporal resources are shared by
allocating certain time slots for each user [7], [8]. Although
in indoor applications, we are concerned with a relatively low
mobility of users, the accurate synchronization required at the
user terminals and at the AP could become a challenge due to
very high data-rates, especially for increased number of users.
Also, in a multi-cell network, TDMA suffers from increased
ICI at overlapping areas between the cells [9]. Obviously, it
can not benefit from the frequency reuse concept in such cases.
B. SDMA
SDMA basically consists in separating users in space domain, which could be realized using angle diversity transmitters (Txs) [9]. In RF systems, SDMA can be achieved using
antenna arrays, which are used for generating narrow beams
pointing at users locations [9]. However, in VLC systems,
this requires employing either special LEDs or using special
optics with the off-the-shelf LEDs. Therefore, from a practical
point of view, SDMA is not well adapted to VLC networks
in general. Nevertheless, it has been shown that combining
TDMA and SDMA can result in a significant improvement in
the network performance, compared with the simple TDMA
approach [9].
C. O-CDMA
By O-CDMA, a unique code, also called signature sequences (SS) is attributed to each user, which is typically an
optical orthogonal code (OOC) [10]. Users then use full time
and spectral resources of the VLC channel. At the receiver
(Rx), each user correlates the received signal with its specific
code in order to (ideally) suppress the IUI. For this, the SS
should be distinguishable from its shifted version as well as
the shifted versions of other users’ SS [4]. Note that random
optical codes (ROCs) are simpler to generate than OOCs [11],
but they provide a sub-optimal performance. In general, to
handle a larger number of users, longer OOCs are needed,
which impacts the achievable data rates and also increases the
system complexity [12]. This can be a serious disadvantage
for a relatively dense VLC network in a typical large-space
scenario.
D. OFDMA
Optical OFDM is a popular transmission technique to
achieve high data-rates in VLC networks, overcoming the
highly constrained modulation bandwidth (BW) of the LEDs
[13].1 When using intensity modulation with direct detection,
which is the case for LED-based links, one drawback of
optical OFDM is the necessity of transmitting a real and
positive signal by, in particular, satisfying the Hermitian
symmetry constraint. This causes a spectral efficiency loss
of factor 2 and 4 for DC-biased optical OFDM (DCOOFDM) and asymmetrically-clipped optical OFDM (ACOOFDM) schemes, respectively [16], [17]. OFDMA relies basically on using an OFDM-based transmission scheme, where
the MA feature is obtained via sharing the spectral resources
(i.e., groups of sub-channels) between users.
For multi-cell scenarios, appropriate frequency reuse solutions can be used to decrease the network ICI [18]. For
instance, fractional frequency reuse (FFR) has been proposed
for DCO-OFDM-based systems in [18], which offers a good
balance between the performance of cell-edge users (CEUs),
the average spectral efficiency, and low system complexity.
Therein, two approaches of strict and soft FFR were considered. Strict FFR is based on dividing the available BW to
1 Note that the LED BW limitation could be circumvented by appropriate
pulse shaping at the Tx, see [14], [15].
a common sub-band, used by cell-center users (CCUs), and
a number of additional sub-bands, used by CEUs; these subbands are assigned so as to minimize the spatial reuse distance.
For soft FFR, CCUs can use the sub-bands reserved to CEUs
in the adjacent cells. In such a case, the performance of CEUs
could degrade due to a higher interference level. To reduce
this effect, more power is allocated to CEUs, compared with
CCUs.
E. NOMA
By power-domain NOMA (that we will simply refer to
as NOMA), multiplexing of users’ signals is carried out at
the AP using superposition coding, which allows the users to
simultaneously use the whole temporal and spectral resources
[19]. NOMA is particularly interesting in VLC applications,
compared for example to the context of 5G RF cellular
networks. Indeed, VLC networks have likely a small number
of users per cell, benefit from relatively high signal-to-noiseratios (SNRs), and typically experience slow channel variations due to the limited user mobility in indoor environments.
At the Rx side, users apply successive interference cancellation (SIC) to its received signal, to partially or fully remove
the IUI from the other users (except for the first user in the
decoding order), requiring hence the corresponding channel
state information (CSI). At the AP, users’ signals are sorted
based on their channel gains, so that users with lower channel
gains are allocated a larger power. These latter decode their
data first at the Rxs side.
F. Comparison of MA Techniques
Following the above mentioned discussions, we take
OFDMA and NOMA schemes as the most relevant for a highrate large-space VLC network. Table I shows a brief comparison between the MA schemes presented above, considering
the three main features of large-space scenarios, i.e., the ICI,
the potential large number of users, and the requirement to
high data rate. We have shown which schemes can support
each of these features.
TABLE I: Comparison between different MA techniques.
TDMA
ICI
Large no. users
High data rate
SDMA
√
√
CDMA
OFDMA
√
√
NOMA
√
√
III. S YSTEM M ODEL AND M ATHEMATICAL F ORMULATION
To characterize the link between the APi and the Rxj , we
define the channel gain hij , which takes into account only the
line-of-sight (LOS) path [20]. Assuming a Lambertian pattern
for the LED luminaire of order m, we have:
hij = S
(m + 1) ρj Aj
cosm (φij ) cos(θij ),
2π ℓ2ij
(1)
where S denotes the LED conversion efficiency, φij the angle
of transmission with respect to APi , θij the angle of incidence
with respect to Rxj , m the Lambertian order of the LED, ρj
the photo-detector (PD) responsivity, and ℓij is the path length
between APi and Rxj . Also, Aj denotes the collection area of
the j th Rx, which is given by:
Aj =
qj2
sin2 (θcj )
APDj ,
(2)
where APDj , θcj , and qj represent the PD active area, the fieldof-view (FOV) of Rxj , and the refractive index of the optical
concentrator, respectively.
For both OFDMA and NOMA approaches, we will consider
signal transmission based on DCO-OFDM.
A. OFDMA-based Signal Transmission
We consider OFDMA with strict FFR approach for ICI
mitigation. We consider determining the BW allocation to
CCUs and CEUs based on the ratio of the number of CCUs
to the total number of users, with respect to the cell which
has the largest number of CEUs, that we denote by ζ, i.e.,
No. of CCUs
.
(3)
ζ=
No. of CCUs + No. of CEUs
This parameter is equal to the fraction of the BW which is
attributed to CCUs:
BWCCU = ζ B,
(4)
where B denotes the total system BW. Note that a different approach was proposed in [18], where a certain non-overlapped
area of the total cell coverage area is considered to include
CEUs (instead of considering CEUs in the intersecting areas
between cells, that we consider in this paper).
Using the approach in (4), the signal-to-interference-plusnoise ratio (SINR) For Rxj is then given by:
p
(hij Pe Kj )2
,
(5)
SINROFDMAj =
σj2
where Pe is the electrical signal power at the AP, Kj denotes
the percentage of power allocated to Rxj , and σj2 stands for
the noise variance. Denoting the allocated BW to Rxj by Bj ,
the maximum achievable rate for Rxj is then:
Bj
log2 (1 + SINROFDMAj ).
(6)
2
B. NOMA-based Signal Transmission
As explained in the previous section, with NOMA, users’
signals are multiplexed in the power domain, while assigning
to each user a power allocation weight, which is smaller for
users with a higher channel gain. At the Rx, to reduce IUI,
SIC is performed, where the decoding order is based on the
channel gain of the users. For instance, the user with the
highest channel gain is the last to decode its signal. Obviously,
for this, the corresponding users’ CSI is required at the Rx.
The received signal at Rxj can be written as [21]:
ROFDMAj =
rj =
j−1
Nr
X
X
p
Pe hij
aik dk + aij dj +
aik dk + zj ,
k=1
k=j+1
(7)
where Pe denotes the transmitted electrical power, dj refers to
Rxj message, and a2ij represents the power allocation weight
of the link from APi to Rxj . In (7), the first term (k < j)
in the parentheses represents the interfering signals that will
be canceled using SIC, the second term contains the desired
signal, and the third term (k > j) represents the residual
interference, which will not be canceled using SIC. Also, zj
represents the additive white Gaussian noise, comprising the
ambient, shot, and thermal noises.
A number of previous works have considered the power
allocation techniques for NOMA signaling such as [22]–[24].
For the sake of simplicity, we consider here the so-called static
power allocation [21], where
a2ij = α a2ij−1 .
(8)
Here, α is the power allocation factor, which represents the
ratio between the allocatedPpowers to successive users in the
Nr
a2ij = 1.
decoding order, such that j=1
For DCO-OFDM based NOMA signal, the upper bound on
the achievable rate for Rxj using NOMA is given by:
B
log2 (1 + SINRNOMAj )
(9)
2
where B represents the signal BW and SINRNOMAj is the
SINR for Rxj :
√
(hij Pe aij )2
SINRNOMAj =
(10)
Nr
X
p
(hij Pe akj )2 + σj2
INOMA +
RNOMAj =
k=j+1
where INOMA denotes the ICI from other cells.
IV. P ERFORMANCE C OMPARISON
We present here numerical results for three different 4cell scenarios to compare the performances of NOMA and
OFDMA. Note that a comparison between these approaches
was presented in [19], [25] but for a single cell architecture.
In the considered scenarios, an LED luminaire serving
as AP handles the Rxs in its cell. A central control unit
connects the APs, and is responsible for exchanging the
channel information of all users between APs, classifying users
into CEUs or CCUs according to the channel information, and
determining the BW allocated to each user. For both OFDMA
and NOMA schemes, CEUs are associated with the AP from
which the receive the strongest signal; this way, the signals
of other APs are considered as interference. This results
in penalizing CEUs in the NOMA approach due to SINR
degradation. By OFDMA, however, CEUs do not suffer from
ICI, because the interfering signals are received on different
frequency bands and, furthermore, every CEU is associated
with the AP corresponding to the highest channel gain. Table
II summarizes the simulation parameters that we consider.2
2 Notice that we have assumed the maximum FOV of 90◦ as we consider
a large room dimension with only 4 APs, so that Rxs can have a LOS link
with the AP at any location inside a cell. For a larger number of APs, i.e.,
reduced cell size, a smaller FOV can be used.
TABLE II: Simulation parameters.
14
(14 m × 14 m × 3 m)
(4.5, 4.5, 2.5)
(4.5, 9.5, 2.5)
(9.5, 4.5, 2.5)
(9.5, 9.5, 2.5)
1
0.4 A/W [19]
1 cm2 [19]
90 deg.
1.5
10 MHz
10− 21 A2 /Hz [19]
7.2 A
0.44 [26]
0.3
12
Rx8
Rx7
10
y-axis (m)
Indoor environment dimension
AP1 location
AP2 location
AP3 location
AP4 location
Lamertian order of LED m
PD responsivity
PD area
FOV of Rx
Refractive index of optical concentrator
System BW B
Noise power spectral density
Current per LED luminaire
LED conversion efficiency S
Power allocation factor α
Rx3
8
Rx2
6
Rx6
Rx5
4
Rx1
Rx4
2
0
0
2
4
6
8
10
12
14
12
14
12
14
x-axis (m)
(a) Top view for Scenario 1
14
12
y-axis (m)
10
Rx12
Rx10
Rx9
Rx11
8
6
4
2
0
0
2
4
6
8
10
x-axis (m)
(b) Top view for Scenario 2
14
12
Rx16
10
y-axis (m)
Figure 1 shows the three considered scenarios with increased complexity from Scenario 1 to 3. Scenario 1 is
composed of 4 CCUs and 4 CEUs such that each AP has
1 CCU and 2 CEUs in its coverage area. For Scenario 2, with
a total of 12 users, we consider 8 CCUs and 4 CEUs, where
each AP has 2 CCUs and 2 CEUs in its coverage area. For
Scenario 3 with a total of 16 users, we consider 12 CCUs and
4 CEUs, while 3 CCUs and 2 CEUs are served by each AP.
All Rxs are considered to be placed at 0.85 m above the floor
level.
Figure 2 presents the maximum achievable throughput for
the users in the three considered scenarios. For OFDMA, we
observe less dissimilarity (or in other words, more homogeneity) in the users’ performance, compared to NOMA. This
results from employing FFR in the OFDMA approach for ICI
mitigation, making CEUs’ performance close to that of CCUs.
Indeed, for OFDMA, users’ performance mostly depends on
their channel gain, the size of the sub-bands allocated to them,
and the power allocated to these sub-bands. On the other
hand, for the NOMA approach, users’ performances largely
depend on their channel gain, their decoding order (which
determines the amount of interference canceled by SIC), and
the power allocation technique used. In fact, although this
power allocation corresponds to the channel gains of the users,
they experience different levels of interference. In addition, for
NOMA, CEUs suffer from ICI because of receiving signals
from more than one AP.
We have further contrasted in Fig. 3 the sum-rate (i.e.,
the sum of maximum achievable throughputs) of CEUs and
the ensemble of users for the two cases of NOMA and
OFDMA signaling and the three considered scenarios of Fig. 1.
Considering the sum-rate of CEUs, we notice that OFDMA
outperforms NOMA in all scenarios, elucidating the efficacy of
FFR in mitigating ICI. Meanwhile, the difference between the
CEUs’ sum-rates with OFDMA and NOMA is less significant
in Scenarios 2 and 3, which shows that increasing the cell
density improves CEUs performance in NOMA, compared
to OFDMA. On the other hand, when considering the total
network sum-rate, we notice that for all scenarios, NOMA
outperforms OFDMA, which shows the merit of the former in
Rx14
8
6
Rx15
Rx13
4
2
0
0
2
4
6
8
10
x-axis (m)
(c) Top view for Scenario 3
Fig. 1: Illustration of the considered scenarios, where × and ◦
refers to APs and users locations, respectively. Scenario 2 has four
additional users with respect to Scenario 1, and Scenario 3 has four
additional users with respect to Scenario 2, whose locations are
indicated on Figs. 1(b) and 1(c), respectively.
OFDMA
NOMA
6
5
4
3
2
1
0
1
2
3
4
5
6
7
Maximum Achievable Throughput (bps)
OFDMA-CEUs
NOMA-CEUs
OFDMA-Total
NOMA-Total
2.5
2
1.5
1
0.5
1
2
3
Fig. 3: Comparison between the sum-rates of the CEUs and the
total network using NOMA and OFDMA approaches for the three
considered scenarios.
7
OFDMA
NOMA
7
6
5
4
3
2
1
0
1
2
3
4
5
6
7
8
9
10
11
12
Receiver
(b) Scenario 2
Maximum Achievable Throughput (bps)
108
Scenario
(a) Scenario 1
10
3
0
8
Receiver
8
3.5
Maximum Achievable Throughput
Maximum Achievable Throughput (bps)
107
7
7
107
OFDMA
NOMA
6
5
4
3
2
1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Receiver
performance for three scenarios of different complexities. We
showed that, for all considered scenarios, NOMA outperforms
OFDMA in terms of the total network sum-rate, which is
similar to the conclusions of the recent literature on singlecell networks. However, we specifically showed that OFDMA
outperforms NOMA in terms of the total achievable throughput of CEUs in a multi-cell network.
Note that, the improvement in network sum-rate using
NOMA comes at the expense of decreased homogeneity of
users’ performance. This results from the difference in the
allocated power and in the interference level experienced
by each user (depending on its decoding order and ICI).
Although OFDMA shows a better performance for CEUs due
to interference mitigation using FFR, increasing user density in
the cells reduces the performance gain with respect to NOMA.
Lastly, regarding the complexity of the MA approach, by
NOMA, the CSI of the other users is required at the Rx side
in order to perform SIC. The quantity of the required CSI
depends on the number of users in a cell and their decoding
order. That is, for increased number of users handled by
an AP, more CSI needs to be transmitted to the users. In
contrast, by OFDMA, ideally the users do not need the CSI
for interference reduction. In this regard, OFDMA can be
considered as less complex to implement.
(c) Scenario 3
Fig. 2: Contrasting users’ maximum achievable throughput using
NOMA and OFDMA approaches for the scenarios presented in Fig. 1.
dealing with multi-cell VLC networks. This clear advantage
can be attributed to partitioning BW resources among users in
OFDMA, whereas in NOMA, users have access to the whole
BW all the time.
V. C ONCLUSIONS
We investigated appropriate MA techniques for large-space
multi-cell VLC networks. Focusing on the most relevant
techniques, i.e., OFDMA and NOMA, we contrasted their
ACKNOWLEDGMENT
This work is supported by the European Union’s Horizon
2020 research and innovation programme under the Marie
Skłodowska-Curie grant agreement no 764461 (VisIoN).
R EFERENCES
[1] Z. Ghassemlooy, L. N. Alves, S. Zvanovec, and M. A. Khalighi, Eds.,
Visible Light Communications: Theory and Applications. CRC-Press,
2017.
[2] S. Al-Ahmadi, O. Maraqa, M. Uysal, and S. M. Sait, “Multi-user
visible light communications: State-of-the-art and future directions,”
IEEE Access, vol. 6, pp. 70 555–70 571, 2018.
[3] H. Marshoud, P. C. Sofotasios, S. Muhaidat, and G. K. Karagiannidis,
“Multi-user techniques in visible light communications: A survey,” in
2016 International Conference on Advanced Communication Systems
and Information Security (ACOSIS), Oct. 2016, pp. 1–6, Marrakesh,
Morocco.
[4] S. S. Bawazir, P. C. Sofotasios, S. Muhaidat, Y. Al-Hammadi, and
G. K. Karagiannidis, “Multiple access for visible light communications:
Research challenges and future trends,” IEEE Access, vol. 6, pp. 26 167–
26 174, 2018.
[5] H. Marshoud, S. Muhaidat, P. C. Sofotasios, S. Hussain, M. A. Imran,
and B. S. Sharif, “Optical non-orthogonal multiple access for visible
light communication,” IEEE Wireless Communications, vol. 25, no. 2,
pp. 82–88, Apr. 2018.
[6] S. Feng, R. Zhang, W. Xu, and L. Hanzo, “Multiple access design
for ultra-dense VLC networks: Orthogonal vs non-orthogonal,” IEEE
Transactions on Communications, vol. 67, no. 3, pp. 2218–2232, Mar.
2019.
[7] J. M. Kahn and J. R. Barry, “Wireless infrared communications,”
Proceedings of the IEEE, vol. 85, no. 2, pp. 265–298, Feb. 1997.
[8] H. Elgala, R. Mesleh, and H. Haas, “Indoor optical wireless communication: potential and state-of-the-art,” IEEE Communications Magazine,
vol. 49, no. 9, pp. 56–62, Sept. 2011.
[9] Z. Chen, D. A. Basnayaka, and H. Haas, “Space division multiple access
for optical attocell network using angle diversity transmitters,” Journal
of Lightwave Technology, vol. 35, no. 11, pp. 2118–2131, June 2017.
[10] J. A. Salehi, “Code division multiple-access techniques in optical fiber
networks - part i: Fundamental principles,” IEEE Transactions on
Communications, vol. 37, no. 8, pp. 824–833, Aug. 1989.
[11] M. F. Guerra-Medina, O. Gonzalez, B. Rojas-Guillama, J. A. MartinGonzalez, F. Delgado, and J. Rabadan, “Ethernet-OCDMA system for
multi-user visible light communications,” Electronics Letters, vol. 48,
no. 4, pp. 227–228, Feb. 2012.
[12] P. H. Pathak, X. Feng, P. Hu, and P. Mohapatra, “Visible light communication, networking, and sensing: A survey, potential and challenges,”
IEEE Communications Surveys Tutorials, vol. 17, no. 4, pp. 2047–2077,
Fourth Quarter 2015.
[13] J. Armstrong, “OFDM for optical communications,” Journal of Lightwave Technology, vol. 27, no. 3, pp. 189–204, Feb. 2009.
[14] S. Hranilovic, “Minimum-bandwidth optical intensity Nyquist pulses,”
IEEE Transactions on Communications, vol. 55, no. 3, pp. 574–583,
Mar. 2007.
[15] W. Gappmair, “On parameter estimation for bandlimited optical intensity channels,” MDPI Computation, Special Issue on Optical Wireless
Communication Systems, vol. 7, no. 1, pp. 1–13, Feb. 2019.
[16] S. Dimitrov and H. Haas, “Information rate of OFDM-based optical
wireless communication systems with nonlinear distortion,” Journal of
Lightwave Technology, vol. 31, no. 6, pp. 918–929, Mar. 2013.
[17] M. A. Khalighi, S. Long, S. Bourennane, and Z. Ghassemlooy, “PAM
and CAP-based transmission schemes for visible-light communications,”
IEEE Access, Special issue on Optical Wireless Technologies for 5G
Communications and Beyond, vol. 5, pp. 27 002–27 013, Oct. 2017.
[18] C. Chen, S. Videv, D. Tsonev, and H. Haas, “Fractional frequency reuse
in DCO-OFDM-based optical attocell networks,” Journal of Lightwave
Technology, vol. 33, no. 19, pp. 3986–4000, Oct. 2015.
[19] L. Yin, W. O. Popoola, X. Wu, and H. Haas, “Performance evaluation of
non-orthogonal multiple access in visible light communication,” IEEE
Transactions on Communications, vol. 64, no. 12, pp. 5162–5175, Dec.
2016.
[20] S. Long, M. A. Khalighi, M. Wolf, S. Bourenanne, and Z. Ghassemlooy, “Investigating channel frequency selectivity in indoor visible light
communication systems,” IET Optoelectronics, vol. 10, no. 3, pp. 80–88,
May 2016.
[21] H. Marshoud, V. M. Kapinas, G. K. Karagiannidis, and S. Muhaidat,
“Non-orthogonal multiple access for visible light communications,”
IEEE Photonics Technology Letters, vol. 28, no. 1, pp. 51–54, Jan. 2016.
[22] C. Chen, W. Zhong, H. Yang, and P. Du, “On the performance of MIMONOMA-based visible light communication systems,” IEEE Photonics
Technology Letters, vol. 30, no. 4, pp. 307–310, Feb. 2018.
[23] Z. Yang, W. Xu, and Y. Li, “Fair non-orthogonal multiple access for
visible light communication downlinks,” IEEE Wireless Communications
Letters, vol. 6, no. 1, pp. 66–69, Feb. 2017.
[24] X. Zhang, Q. Gao, C. Gong, and Z. Xu, “User grouping and power
allocation for NOMA visible light communication multi-cell networks,”
IEEE Communications Letters, vol. 21, no. 4, pp. 777–780, Apr. 2017.
[25] R. C. Kizilirmak, C. R. Rowell, and M. Uysal, “Non-orthogonal multiple
access (NOMA) for indoor visible light communications,” in International Workshop on Optical Wireless Communications (IWOW), Sept.
2015, pp. 98–101, Istanbul, Turkey.
[26] H. Ma, L. Lampe, and S. Hranilovic, “Coordinated broadcasting for multiuser indoor visible light communication systems,” IEEE Transactions
on Communications, vol. 63, no. 9, pp. 3313–3324, Sept. 2015.