Power Domain Sparse Dimensional Constellation

Multiple Access (PD-SDCMA) for Enabled Flexible

PONs
Yuhao Lian∗§ , Xiao Han∗† and Xinmao Deng∗†
† Chongqing University of Posts and Telecommunications, Chongqing 400065, China
§ College
of Engineering, Zhejiang University, Hangzhou 310027, China
Email: [email protected], [email protected], yuhao [email protected]
∗ These authors contributed equally to this work
arXiv:2506.08053v1 [cs.ET] 9 Jun 2025

Abstract—With the commercial deployment of 5G and the in- cost-effective TWDM-PON [9], and evolves to OFDM-PON,
depth research of 6G, the demand for high-speed data services offering higher speeds and greater tolerance to dispersion [10]
in the next-generation fiber optic access systems is growing [11]. OMA-PON maximizes high-speed and high-capacity
increasingly. Passive optical networks (PONs) have become a
research hotspot due to their characteristics of low loss, high user requirements by utilizing orthogonal resources.
bandwidth, and low cost. However, the traditional orthogonal However, the continuous advancement of PON standards
multiple access (OMA-PON) has difficulty meeting the require- has elevated the requirements for next-generation PON system
ments of the next-generation PON for high spectral efficiency and architecture, such as the standardization of NG-PON2
and flexibility. In this paper, a novel transmission technology, allows a differential ROP of 15 dB in each optical device
namely power-domain sparse dimension constellation multiple
access (PD-SDCMA), is proposed for the first time. Through the [12]. The traditional OMA-PON are no longer sufficient to
signal space dimension selection strategy (S2D-strategy) in the meet these requirements. To accomplish such a challenging
high-dimensional signal space, the low-dimensional constellation task cost-effectively, the multi-stack architecture of Flexible
is sparsely superimposed into the high-dimensional space, thereby PON based on PD-NOMA is considered a promising candidate
reducing multi-user interference and enhancing the system ca- due to its high spectral efficiency and scalability [13] [14]. PD-
pacity. PD-SDCMA supports higher-order modulation formats
and more access groups, and is also compatible with the existing NOMA adapts to the path loss of different access groups by
orthogonal frequency division multiplexing (OFDM) architecture. providing distinct power budgets, allowing users from all ac-
The simulation results show that in a 25 km single-mode fiber cess groups to be controlled and optimized collectively through
system, compared with PD-NOMA and 3D-NOMA, PD-SDCMA digital means, thereby enhancing overall service quality [15].
can support more users and significantly reduce BER. This Additionally, Flexible PON based on PD-NOMA demonstrates
technology provides an efficient and low-cost solution for the
evolution of Flexible PONs. robustness in power distribution parameters, adapting to un-
Index Terms—PD-SDCMA, S2D-strategy, OMA-PON, Flexible certainties caused by temperature variations and aging in fiber
PONs networks by adjusting the power distribution. [16].
It is widely acknowledged that to further enhance the
I. I NTRODUCTION transmission capacity and correspond-ding spectral efficiency
As the commercial deployment of 5G continues to progress of Flexible PON based on PD-NOMA, the adoption of higher-
and research efforts into 6G gradually deepen, there is an order signal modulation formats and an increase in the number
escalating demand for High-speed data service in the next of access groups is necessary. However, this approach brings
generation of fiber optic access systems [1] [2]. In this context, about 2 significant drawbacks: a) the high serial interference
Passive Optical Network (PON) have garnered widespread at- of PD-NOMA cannot meet the demands of higher-order mod-
tention due to their low-loss characteristics, ample bandwidth, ulation formats and increased access group numbers; and b)
and cost-effectiveness in optical access network [3]. Currently, as the number of access groups increases, the power control
A suite of standards Recommendati-ons for a 50 Gb/s line rate range for individual access groups becomes relatively smaller.
PON system has been developed by the ITU-T, marking a This leads to the need for enhanced sensitivity in power control
significant leap from the current 10 Gb/s deployments in fiber devices on the technical level, thereby resulting in increased
access applications [4]. However, achieving this substantial network deployment and management costs.
increase in performance requires a corresponding evolution To avoid these unnecessary drawbacks and adapt to higher
in the underlying multiple access technologies. Hence, the signal modulation formats and enhance the information-
underlying technology of PON system is continuously evolv- carrying capacity of individual access groups, PSCD-NOMA
ing. It begins with the straightforward and easy-to-manage has been proposed [17]. This technique employs sparse coding,
TDM-PON [5] [6], progresses to the more scalable WDM- treating the information of each user access group as a
PON [7] [8], then advances to the more flexible and more new constellation. Specifically, an IS-8QAM constellation is
designed based on optimal MED and Constellation Figure of obtained from the simulations and the main contributions of
Merit (CFM). Theoretical analysis shows that, under the same the article.
power level, using sparse coding constellations can achieve
a 150% overload, meeting the architectural characteristics II. PD-SDCMA OPERATING PRINCIPLE
of Flexible PON while serving more users. Inspired by the
fundamental concept of 3D-OFDM, research recently has A. Vector space representation of signals
proposed 3D-NOMA [18]. This approach enhances the MED
of constellation points by introducing 3D constellation, facili-
tating easier demodulation of user information and increasing
the number of access groups. Theoretical results indicate that,
compared to the QPSK constellation, the average transmission
power of the 3D tetrahedron constellation points in 3D-NOMA
can be reduced by 13.07%, while the MED is increased by
15.48% under the same power. Experimental results show
that 3D-NOMA allows for more access groups, with a higher
fluctuation tolerance of the Received Optical Power (ROP).
This enhances the robustness of power distribution parameters
in Flexible PON, reduces the demand for power control
devices, and ultimately lowers both the technical and network
deployment costs.
For the first time, this paper introduces a novel transmission
technology known as Power Domain Sparse Dimensional Con-
stellation Multiple Access (PD-SDCMA) for Flexible PON.
PD-SDCMA employs a signal space dimension selection
strategy (S2D-strategy) based on the superposition of high-
dimensional spaces, utilizing low-dimensional constellations
to achieve the superposition of sparse constellation points
within the high-dimensional constellation space. By leverag-
ing sparse, unrelated dimensions, PD-SDCMA reduces the
high serial interference caused by Superposition Coding (SC),
thereby mitigating the error propagation effect [19]. The main
contributions of this paper are summarized as follows: Fig. 1. Schematic illustration of S2D-strategy.

• PD-SDCMA scheme significantly enhanced the accept- In the vector space analysis theory of digital signals, a
able constellation modulation order and the number of point in the vector space represents a signal waveform. The
coupled access groups. This improvement stems from its orthonormal basis functions in the vector space correspond to
signal space dimension selection strategy based on high- the respective spatial dimensions. If N functions fk (t), where
dimensional space superposition, which employs low- k = 1, 2, ..., N , satisfy:
dimensional constellations to achieve sparse constellation
point superposition in a high-dimensional signal space, ∞
(
m ̸= n
Z
0,
thereby reducing interference between access groups. fn (t)fm (t) = (1)
• Moreover, we offered a design scheme compatible with −∞ 1, m=n
any constellation and retained all the benefits of OFDM.
Moreover, developing PD-SDCMA based on higher-
dimensional constellation design schemes will be a sig-
nificant direction for future research.
This paper is organized as follows: In Section II, we
present a comprehensive description of the PD-SDCMA sys-
tem model, where particular emphasis is given to theory of
the signal space dimension selection strategy (S2D-strategy)
and sparse superposition method for PD-SDCMA from the
perspective of the vector space representation of signals. Then,
in Section III, the relevant simulation parameters are provided,
we perform simulations to compare the supporting access users
with PD-NOMA and 3D-NOMA in a 25 km SMF system. Fi-
nally, Section IV concludes the paper, summarizing the results Fig. 2. The DSP processing procedure at the transmitter.
 Fig. 3. Schematic illustration of S2D-strategy to constellation.

The functions are orthogonal to each other, and the function dimension. Therefore, the traditional two-dimensional QPSK
set {fk (t), k = 1, 2, · · · , N } is called an orthonormal basis constellation can be represented in a four-dimensional signal
function set. Suppose a signal waveform S(t) can be repre- space as:
sented as a linear combination of fk (t), that is:
N
S = [s1 , s2 , 0, 0], s1 , s2 ∈ {±1} (6)
X
S(t) = sk fk (t) (2) B. S2D-strategy for sparse superposition
k=1
Fig. 1 presents different constellation superposition methods
The position of a signal in the vector space can be deter-
in the power domain. The horizontal axis represents power, and
mined by the projections of the signal waveform S(t) onto
the vertical axis represents dimensions. Different rectangular
{fk (t), k = 1, 2, · · · , N }. Therefore, the signal waveform
columns correspond to different user access groups in the
S(t) can be represented in an N -dimensional vector space
Flexible PON. The arrangement of the rectangular columns
as:
along the horizontal axis indicates the superposition of the
constellations of the access groups in the power domain. In
S = [s1 , s2 , · · · , sN ] (3)
Fig. 1, a rectangular column composed of multiple squares
In signal analysis, the above-mentioned N -dimensional vec- indicates that the constellation of the access group can be
tor space is called an N -dimensional signal space. The M mapped to the N dimensions of the N -dimensional signal
points in the N -dimensional signal space onto which the signal space. The squares marked with ”✓” in the rectangular col-
waveform is mapped are called the constellation of an M -ary umn represent the signal space dimensions allocated to the
signal. In order to represent a signal space with more than constellation of this access group. Therefore, it can be seen
two dimensions, the linear combination of S(t) must contain from Fig. 1(a) that the constellations of the access groups
more than two orthonormal basis functions. When a signal in PD-NOMA only perform continuous power superposition
waveform is mapped into a signal space, the in-phase and in two dimensions of the signal space. In Fig. 1(b), when
quadrature components of a carrier at the same frequency can considering an Flexible PON with three access groups, the
only be represented as two orthonormal basis functions, that is, N -dimensional signal space S is divided into N/3 subspaces
two dimensions in the signal space. Therefore, a signal space Si (i = 1, 2, ..., N/3), where Si contains three signal space
with more than two dimensions is composed of the in-phase dimensions. Thus, S1 contains the first three dimensions of
and quadrature components of multiple orthogonal carriers. the N -dimensional signal space, S2 contains the fourth to
The orthogonal carriers must satisfy: sixth dimensions of the N -dimensional signal space, and S3
Z Ts contains the seventh to ninth dimensions of the N -dimensional
cos(2πfi t + φi ) cos(2πfj t + φj )dt = 0 (4) signal space. Therefore, the relationship between the mth
0 dimension of the subspace Si and the Mth dimension of the
where Ts is the symbol-period time, fi and φi are the signal space is:
frequency and phase of the ith carrier, respectively, and fj and
φj are the frequency and phase of the jth carrier, respectively. m = M − (i − 1)P (7)
Based on this, if the traditional two-dimensional orthogonal
where P represents the dimension of the signal space con-
constellation is mapped into an N -dimensional signal space,
tained in the subspace Si , and P ≥ g as well as P ≥ 2.
the signal waveform S(t) can be expressed as:
As shown in Fig.1(b), the constellation of the first access
group is mapped to the first and second dimensions of each
S = [s1 , s2 , 0, 0, · · · , 0] (5)
subspace, the constellation of the second access group is
where S contains N − 2 zeros, corresponding to the in-phase mapped to the second and third dimensions of each subspace,
and quadrature components of the orthogonal carriers in that and the constellation of the third access group is mapped to
the third and first dimensions of each subspace. Therefore,
in the first dimension of each subspace, there is only the 0 −1
NX
power superposition of the constellations of the first access S(t) = Si (t)
group and the third access group. In the second dimension of i=0
each subspace, there is only the power superposition of the N0 /Ns −1 Ns −1
X X
constellations of the first access group and the second access = Re{[Iik (t) + jQik (t)]ej2πfik t }
group. In the third dimension of each subspace, there is only (10)
i=0 k=0
the power superposition of the constellations of the second (N −1
0
X
)
access group and the third access group. The above discrete = Re [In (t) + jQn (t)]ej2πn∆f t ej2πf0 t
superposition method of constellations of different access n=0
groups in the power domain is called constellation sparse = SI (t) cos(2πf0 t) − SQ (t) sin(2πf0 t)
power superposition. the strategy in which constellations with
different power proportions select the mapping of specific where N0 is the number of orthogonal carriers corresponding
dimensions of each subspace is called the S2D-strategy. At to the N -dimensional signal space, ∆f is the interval of
this time, the S2D-strategy can be represented by a matrix orthogonal frequencies. In (t) and Qn (t) are the in-phase and
Sg×2 with g rows and 2 columns: quadrature components of the nth carrier corresponding to
the mappingPNdimensions in the N -dimensional signal space.
0 −1
S(t) = [I (t) + jQn (t)ej2πn∆f t ] is the complex-
 
1 2 b
n=0 n
 2 3 envelope signal. SI (t) and SQ (t) are the real and imaginary
 
Sg×2 =  · · ·
 · · · (8) parts of the complex-envelope signal. Within a symbol interval

g − 1 g  Ts , N0 equally-spaced samples are taken for the complex-
g 1 envelope signal S(t), and the sampling interval is Ts /N0 , that
is:
where g rows represent the g access groups of the Flexible
PON, and the two columns represent the two dimensions in
0 −1
NX
the signal subspace to which the constellation of each access
group is mapped. The ith row vector represents the dimensions S(kTs /N0 ) = [I(n) + jQ(n)]ej2πn∆f kTs /N0
n=0
in the P -dimensional subspace to which the constellation of NX0 −1
the ith access group is mapped. For example, the first row
= [I(n) + jQ(n)]ej2πnk/N (11)
vector indicates that the dimensions in the P -dimensional
n=0
subspace to which the constellation of the first access group NX0 −1
is mapped are the first and second dimensions, that is, the in- = X(n)ej2πnk/N
phase component and the quadrature component of the same n=0
frequency carrier. One P -dimensional subspace corresponds to
the in-phase components and quadrature components of mul- The above formula is the IDFT of the discrete sequence.
tiple orthogonal carriers. Therefore, in the ith P -dimensional Therefore, the transmitter generates the signal by using the
subspace with the number of orthogonal carriers being , the IFFT operation, and the receiver can recover the signal by
signal Si (t) can be expressed as: adopting the FFT.
The DSP processing flow at the transmitting end is shown
in Fig.2. First, the high-speed serial bit-stream is converted
s −1
NX
into multiple sets of parallel data through a Serial-Parallel
Si (t) = Iik (t) cos(2πfik t) − Qik (t) sin(2πfik t)
Converter (S/P conversion). Subsequently, these multiple sets
k=0
(9) of parallel data undergo constellation mapping to become
NXs −1

= Re{[Iik (t) + jQik (t)]ej2πfik t } symbols. According to the S2D-strategy, the symbols are
k=0
mapped to the corresponding signal space dimensions, and the
real/imaginary parts corresponding to the unmapped subspace
where fik represents the frequency of the kth orthogonal dimensions are left vacant, which is the symbol reconstruction
carrier corresponding to the ith subspace. Qik (t) and Iik (t) process. After completing symbol reconstruction, an IFFT
represent the in-phase and quadrature components of the kth operation is performed on the reconstructed symbols.
orthogonal carrier corresponding to the dimensions mapped
by the S2D-strategy in the ith subspace. Here, the in-phase
N
and quadrature components corresponding to the subspace 1 X
Y = IF F T {X} → Y (n) = X(k)WN−nk (12)
dimensions to which the constellation is mapped are non-zero, N
k=1
while the remaining quadrature and in-phase components are
set to zero. Therefore, the signal S(t) generated corresponding A Cyclic Prefix (CP) is inserted into the generated time-
to the combination of subspaces into an N -dimensional signal domain signal, and then a Parallel-Serial (P-S) conversion is
space is: carried out.
 Fig. 4. The simulation platform and electrical signal spectrum.

TABLE I
PARAMETER S ETTINGS
′
Y = [y0′ , y1′ , y2′ , · · · ′
, yN ′ ′ ′
−2 , yN −1 , yCP 0 , yCP 1 , · · · ] (13)
Parameter Value
Finally, the g signals generated by the g access groups Constellation QPSK & 16QAM
are summed in the power-domain to generate the transmitted Number of orthogonal carriers 256
IFFT/FFT points 512
signal. Number of symbols 1000
CP 0.125
p p Electrical signal bandwidth 10Gb/s
P1 Y1′ (t) + P2 Y2′ (t) + · · · + Pg Yg′ (t)
p
Z(t) = (14) Laser center wavelength 1550nm
Fiber attenuation 0.2dB/km
P1 + P2 + · · · + Pg = 1 (15) Fiber dispersion 16ps/(nm · km)

At the receiving end, Successive Interference Cancellation

(SIC) can separate and demodulate signals successively. Un- TABLE II
S2D- STRATEGY
like PD-NOMA’s SIC demodulation, PD-SDCMA demodu-
lation maps the superimposed joint constellation onto the Scenarios Constellation S2D-strategy Power distribution
accessing user’s constellation plane.As in Fig.3, consider the 2 groups 16QAM [1 2;2 3] 16:1
sparse superposition of QPSK constellations of two access 3 groups QPSK [1 2;2 3;3 1] 16:4:1
5 groups QPSK [1 2;2 3;3 4;4 5;5 1] 256:64:16:4:1
groups. The transmitter gives high power to the first-access
user and low power to the second. In 3D space, the first-access
user’s four constellation points are on the xoy plane, and the the optical receiver module. The S2D-strategy, parameters
second-access user’s constellation is on the yoz plane. Their of important optoelectronic devices, and system simulation
superposition forms a 16-point cuboid joint constellation in parameters are shown in Tables I and II.
the signal space.During demodulation, the receiver treats the
low-power group as noise to demodulate the high - power B. RESULTS ANALYSIS
one, mapping 16 points onto the xoy plane. Since the first- Fig. 5(a) shows the relationship curve between BER and
access user’s constellation has no z-axis mapping, points the Receiving Optical Power (ROP) corresponding to the
with the same xy-coordinates coincide on the xoy plane, scenario of two access groups under 16QAM constellation
transforming the joint constellation into an 8QAM-like one. mapping. As can be seen from Fig. 5(a), under the HD-
Serial interference on the z-axis doesn’t affect the first-access FEC threshold (3.8 × 10−3 ), PD-SDCMA can achieve re-
user’s demodulation, and the first-access user has less impact liable 16QAM transmission, which cannot be realized by
on the second-access user’s demodulation. PD-NOMA. Meanwhile, under the condition of using QPSK
constellation mapping, Fig. 5(b) shows BER and ROP curves
III. SIMULATION AND RESULTS ANALYSIS
for the scenario with three access groups. In order to conduct a
A. Experimental Settings performance comparison with traditional PD-NOMA and 3D-
In this section, the Flexible PON simulation transmission NOMA, the simulation results of the three NOMA schemes
platform based on PD-NOMA, 3D-NOMA and PD-SDCMA under the same simulation conditions are presented in Fig.
is built using Matlab 2020a and VPI TransmissionMaker 5(b). In Fig. 5(b), PD-NOMA can no longer ensure that the
10.0. Matlab mainly implements the DSP functions in the BER performance of the three access groups reaches the HD-
transceiver, while VPI accurately conducts numerical model- FEC threshold, indicating that the upper limit of the number of
ing of various optoelectronic devices. As shown in Fig. 4, access groups for PD-NOMA PON is two. When comparing
the simulation platform consists of three parts: the optical the 3D-NOMA and PD-SDCMA schemes, under the HD-FEC
transmitter module, the optical fiber transmission link, and threshold, the minimum received sensitivities of 3D-NOMA
 Fig. 5. BER curves of different multiple-access schemes.

and PD-SDCMA are approximately -10.2 dBm and -10.8 dBm [5] K. L. Hall, “High-speed TDMA techniques,” Optical Fiber Communi-
respectively. Therefore, in the case of three access groups, cation Conference., vol. 3, pp. 311–312, Nov. 2002.
[6] L. G. Kazovsky, W.-T. Shaw, Dimitri Gutiérrez, N. Cheng, and S.-
PD-SDCMA can increase the optical power budget of the W. Wong, “Next-Generation Optical Access Networks,” Journal of
traditional PD-NOMA-based Flexible PON transmission by Lightwave Technology, vol. 25, no. 11, pp. 3428–3442, Nov. 2007.
0.6 dB. To evaluate the access flexibility of PD-SDCMA, [7] S.-J. . Park, C.-H. . Lee, K.-T. . Jeong, H.-J. . Park, J.-G. . Ahn,
and K.-H. . Song, “Fiber-to-the-Home Services Based on Wavelength-
the simulation system takes into account the Flexible PON Division-Multiplexing Passive Optical Network,” Journal of Lightwave
application scenario with five access groups. As shown in Fig. Technology, vol. 22, no. 11, pp. 2582–2591, Nov. 2004.
5(c), the increase in the number of access groups leads to [8] K. Grobe and J.-P. . Elbers, “PON in adolescence: from TDMA to
a decrease in the constellation MED. Even with the QPSK WDM-PON,” IEEE Communications Magazine, vol. 46, no. 1, pp.
26–34, Jan. 2008.
constellation mapping, 3D-NOMA cannot ensure that BER [9] Z. Tan, C. Yang, and Z. Wang, “Energy Consume Analysis for Ring-
performance of each access group meets the requirements of Topology TWDM-PON Front-Haul Enabled Cloud RAN,” Journal of
the HD-FEC threshold. Lightwave Technology, vol. 35, no. 20, pp. 4526–4534, Oct. 2017.
[10] J. Armstrong, “OFDM for Optical Communications,” Journal of Light-
wave Technology, vol. 27, no. 3, pp. 189–204, Feb. 2009.
IV. C ONCLUSION
[11] C. Chow, Chien Hung Yeh, C. H. Wang, C.-Y. Wu, S. Chi, and C.
The PD-SDCMA technology proposed in this paper ef- Lin, “Studies of OFDM signal for broadband optical access networks,”
IEEE Journal on Selected Areas in Communications, vol. 28, no. 6, pp.
fectively addresses the challenges of interference and power 800–807, Aug. 2010.
control in traditional PD-NOMA through the sparse dimension [12] L. Yin, Y. Zhu, Q. Wu, H. Jiang, Qunbi Zhuge, and W. Hu, “Pilot-Tone
allocation strategy in the high-dimensional signal space. Sim- Assisted Successive Interference Cancellation for Uplink Power- and
Frequency-Division Multiplexing Passive Optical Network,” Journal of
ulation results demonstrate that PD-SDCMA significantly out- Lightwave Technology, vol. 40, no. 13, pp. 4237–4245, Mar. 2022.
performs existing schemes in terms of user capacity, and power [13] Y. Saito, Y. Kishiyama, A. Benjebbour, T. Nakamura, A. Li, and K.
sensitivity, and it is particularly suitable for Flexible PON Higuchi, “Non-Orthogonal Multiple Access (NOMA) for Cellular Future
scenarios featuring high density and high dynamics. Future Radio Access,” 2013 IEEE 77th Vehicular Technology Conference (VTC
Spring), Jun. 2013.
work will focus on the optimized design of multi-dimensional [14] F. Lu, M. Xu, L. Cheng, J. Wang, and G.-K. Chang, “Power-Division
constellations and the integration with artificial intelligence- Non-Orthogonal Multiple Access (NOMA) in Flexible Optical Access
driven dynamic resource allocation to further enhance the sys- With Synchronized Downlink/Asynchronous Uplink,” Journal of Light-
wave Technology, vol. 35, no. 19, pp. 4145–4152, Oct. 2017.
tem performance. PD-SDCMA offers an innovative pathway [15] L. Dai, B. Wang, Y. Yuan, S. Han, C. I, and Z. Wang, “Non-orthogonal
for the low-cost and high-spectral-efficiency evolution of next- multiple access for 5G: solutions, challenges, opportunities, and future
generation fiber optic access networks. research trends,” IEEE Communications Magazine, vol. 53, no. 9, pp.
74–81, Sep. 2015.
R EFERENCES [16] Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, and V.
K. Bhargava, “A Survey on Non-Orthogonal Multiple Access for 5G
[1] H. Tataria, M. Shafi, A. F. Molisch, M. Dohler, H. Sjoland, and F. Networks: Research Challenges and Future Trends,” IEEE Journal on
Tufvesson, “6G Wireless Systems: Vision, Requirements, Challenges, Selected Areas in Communications, vol. 35, no. 10, pp. 2181–2195,
Insights, and Opportunities,” Proceedings of the IEEE, vol. 109, no. 7, Oct. 2017.
pp. 1166–1199, Jul. 2021. [17] S. Chen et al., “Power Sparse Code Division Non-Orthogonal Multiple
[2] T. S. Rappaport et al., “Wireless Communications and Applications Access Scheme for Next-Generation Flexible Optical Access,” Journal
Above 100 GHz: Opportunities and Challenges for 6G and Beyond,” of Lightwave Technology, vol. 40, no. 22, pp. 7236–7245, Aug. 2022.
IEEE Access, vol. 7, pp. 78729–78757, 2019. [18] Chen S et al., “Performance improvement of non-orthogonal multiple
[3] Jun Shan Wey, “The Outlook for PON Standardization: A Tutorial,” access with a 3D constellation and a 2D IFFT modulator,” Optics
Journal of Lightwave Technology, vol. 38, no. 1, pp. 31–42, Jan. 2020. Express, vol. 31, no. 5, pp. 7425–7439, 2023.
[4] R. Bonk et al., “50G-PON: The First ITU-T Higher-Speed PON Sys- [19] H. Li, Z. Huang, Y. Xiao, S. Zhan, and Y. Ji, “Solution for error
tem,” IEEE Communications Magazine, vol. 60, no. 3, pp. 48–54, Mar. propagation in a NOMA-based VLC network: symmetric superposition
2022. coding,” Optics Express, vol. 25, no. 24, pp. 29856–29856, Nov. 2017.