High-Capacity Free Space Optics-Based Passive Optical Network for 5G Front-Haul Deployment

Abstract

With the expansion of Information and Communication Technology, it is important to develop a communication network that can provide high-capacity ubiquitous connectivity. This work proposes an energy-efficient passive optical network (PON) using orthogonal frequency division multiple access (OFDMA) and wavelength division multiplexing (WDM) to facilitate the dense deployment of radio units (RUs) in a beyond 5G (B5G) communication network. High-speed connectivity is ensured by employing a hybrid PON architecture that includes a combination of free space optics (FSO) links and optical fiber (OF) media to carry OFDM and WDM multiplexed traffic. Furthermore, an optical frequency comb generator (OFCG) is utilized at the transmitter module to generate and leverage the spectrum for transmitting information from baseband units (BBUs) to the RUs situated near the end users. The proposed system is analyzed through (i) simulation analysis using Optisystem for transmission capacity computations and (ii) mathematical analysis to determine the total savings in energy. The simulation analysis shows that the given architecture can carry data across 3 km of FSO medium using 512 subcarriers per BBU transmitting at 10 Gbps of data with QPSK-modulated bit sequence. Additionally, energy efficiency shows that the use of an OFCG cuts the total energy usage by 22% at the transmitter module without negatively impacting the system’s high cardinality and transmission capacity.

1. Introduction

The widespread growth of data communication devices has considerably increased over the past two decades to improve the living standard of the masses. However, the construction of densely interconnected networks necessitates considerable bandwidth capacities due to integrating numerous devices into a restricted space. Digital subscribers’ line (DSL) and optical fiber (OF) media, two examples of traditional wired networks, can provide the necessary high capacity. They do not, however, immediately link wireless devices at the front end.

To connect a high number of devices within a cell, wireless networks, like 3G and 4G, rely heavily on base stations with wireless connectivity. However, these networks cannot handle the increasing number of linked devices and the bandwidth they need [1,2,3,4,5]. Furthermore, regular radio communication links, like radio frequency (RF) and microwaves, are interference limited. This problem in micro and millimeter waves reduces their capability to support a large cardinality front-haul network [6]. Therefore, it is important to utilize a network that can interconnect a large number of heterogeneous devices with the promise of high-capacity communication in terms of latency, data rate, cardinality, and distance between the transmitter and receiver modules.

From the 2G era to the advent of 4G, the primary deployment approach revolved around the distributed radio access network (DRAN) model. Within DRAN, the cell site hosted both the baseband unit (BBU) and the remote radio head (RRH) [7]. With the advent of 5G and beyond, the deployment paradigm adopted a centralized structure known as a cloud radio access network (CRAN). The fundamental configuration of CRAN comprises three key elements: a baseband unit (BBU) pool housing numerous BBUs equipped with centralized processors, radio units (RUs) housing antennas positioned at the cell site, and a front-haul network linking RUs and BBUs through high-capacity, low-latency connections [7]. The design of the front-haul network to provide a high-capacity, low latency network with cost and energy efficiency has been one of the most significant design challenges on the road to 5G, beyond 5G (B5G), and 6G realization because these networks require dense deployment of RUs near the subscriber premises [2,3,4,5,6,7].

In CRAN architecture, the base station is divided into BBUs and RUs. The BBUs are typically placed at a centralized location, whereas the RUs are deployed at the cell site, with limited functionalities in the lower layer split (LLS). Both the BBUs and RUs are connected via a front-haul network, as shown in Figure 1 [8]. Within this setup, numerous BBU nodes efficiently distribute resources to RUs, adjusting to the network’s real-time demands. Meanwhile, an RU, along with an antenna mounted on the cell tower (CT), is used to establish connections with wireless devices.

The widespread installation of RUs introduces significant design complexities to the front-hauling process. This arises from the need for a robust communication channel capable of carrying high-capacity traffic with minimum latency between the transmitter and receiver modules at the BBU and RUs [1,3]. One approach to overcome this problem is deploying an all-fiber path installed between the BBU pools at the CRAN module and their corresponding RUs to support high capacity and low latency in B5G and 6G networks. However, this solution comes at a considerable cost, with deployment expenses soaring 10 times due to the extensive civil work required for trenching and planning, particularly in densely populated urban areas [3,9].

Consequently, this paper proposes a hybrid passive optical network (PON) that combines a free space optics (FSO) link and optical fiber media to deploy a high-speed and low-cost interconnection between the BBU pool and the corresponding RUs. FSO is a high-speed communication technology that facilitates wireless data transmission by transmitting a narrow spectrum (often produced by laser diodes (LDs)) through the air, outer space, or a vacuum.

However, the performance of the FSO link is subject to degradation by the effects of the link and weather attenuations, along with random variation in the signal intensity and phase caused by atmospheric turbulence. This occurrence bears similarities to multipath fading observed in radio frequency wireless communication. It arises from the stochastic changes in the atmosphere’s refractive index due to fluctuations in temperature and pressure [10]. Therefore, it is important to adapt suitable communication technology that can fulfill the requirements of the dense deployment B5G and provide high data rates at low latency in an FSO-enabled PON in severe conditions.

Different techniques have been investigated for deploying a high-capacity, low-latency path that can provide the desired bandwidth with high cardinality support. A Quadrature amplitude modulation (QAM)-based orthogonal frequency division multiplexing (OFDM) system is proposed in [11], where orthogonal band multiplexing is utilized for traffic aggregation to support data rates of up to 10 Gbps. This architecture utilizes an all-fiber path without the application of FSO links. Furthermore, amplifiers are not utilized to ensure energy efficiency, limiting receiver sensitivity to −17 dBm. Chung Yi-Li et al. proposed a wavelength division multiplexing (WDM)-based FSO-PON to support 10 Gbps data down- and up-link transmission for 5G communication networks [12]. They used a vertical cavity surface emitting laser (VCSEL) to implement their architecture. Their analysis shows they can achieve receiver sensitivity of −16 dBm for an acceptable BER at a 600 m FSO link. Furthermore, the application of an EDFA, temperature controller, VCSEL-based wavelength selector yields poor energy efficiency for application at the access domain.

An OFDM-based model is proposed in [13] to deploy the RF front-end for 5G networks. A photonic-based RF converter (PBRC) is employed to convert low frequencies into RF frequencies that essentially contain a continuous wave (CW) laser diode, MZM, and a photodetector arrangement. They can achieve suitable data rates for 2G, 3G, 4G, and 5G communication. However, the application of PBRC will result in poor energy efficiency. Junyi Zhou in [14] proposes a 10 Gbps 16 PSK-based OFDM-FSO system using phase difference. The proposed system can support communication over 500 m FSO links. Furthermore, multiple electrical gain components are used at the transmitter module, which can result in poor energy efficiency for the system.

C. H. Yeh et al. proposed a PON formed by combining an optical fiber and an FSO link at the ODN level [15]. The proposed model can support 10 Gbps of data with on–off keying modulation over a 300 m FSO link at a receiver sensitivity of −20 dBm. In [16], the authors have proposed a 4 × 10 Gbps downlink model with the application of OFDM and a fiber–FSO link combination at the ODN level. The proposed work is able to achieve 5G-compatible data rates for up to 650 m FSO links. Nahal and Xu [17] have used a WDM-PON to transmit OFDM-based signals for 5G communication networks. The proposed model can achieve 20 Gbps downstream transmission over a 20 km SMF and a 500 m FSO link. Energy efficiency is applied by using a wavelength reuse scheme based on wavelength-seeded RSOAs.

In the quest to deploy a low-cost, high-capacity PON with minimum latency, this work utilizes a combination of WDM and OFDMA to provide the required connectivity between the BBU pool and RUs for 5G and beyond and 6G networks. OFDMA facilitates connectivity for many subscribers over the wireless medium. At the same time, WDM ensures smooth traffic integration from multiple BBUs in the pool. Furthermore, the CRAN arrangement utilizes an optical frequency comb generator (OFCG) to generate the required spectrum. The proposed setup, with the utilization of an OFCG module for WDM modulation at the transmitter and FSO links at the distribution level (DL), will ensure high-capacity transmission and enable an energy-efficient architecture with minimum capital expenditure (CAPEX).

Performance analysis is performed regarding the transmitter module’s bit error rate (BER) and total energy consumption. BER analysis is performed by developing a realistic simulation model, Optisystem, to validate the claim for network compatibility to support high capacity and cardinality for B5G and 6G communication models. Furthermore, energy consumption is determined through mathematical formalism and compared with the conventional architecture to determine the improvement in overall energy consumption through the proposed changes.

2. Proposed Architecture

Figure 2 shows the proposed hybrid architecture for interconnection between the BBU pools at CRAN and their corresponding RU near the subscriber premises. The BBU pool is formed through the combination of multiple BBUs along with the OFCG source. Each BBU is constructed through a pseudo-random bit source (PRBS) fed into a QPSK sequence generator for initial modulation with two symbols per bit. Serial data from the QPSK modulator are given at the OFDM modulator’s input port, as demonstrated in Figure 3. The OFDM modulation’s initial step involves converting the serial symbols stream from the QPSK modulator into a parallel format using a serial-to-parallel converter. This transformation is vital as it enables the transmission of more information than serial communication. The output from the serial-to-parallel converter is then fed into an inverse Fourier transform (IFFT) block, as illustrated in Figure 3, to determine the corresponding waveform in the time domain.

For further efficiency, the output of the IFFT block is fed to an Add Cyclic Prefix unit that introduces a guard period at the start of each symbol to overcome the adjacent interference in the time domain. A digital to analog (DAC) converter is employed after the cyclic prefix block, which converts the parallel data streams into serial streams for efficient transmission over the communication channel.

WDM modulation is performed with the help of an OFCG arrangement and RF-up-converter, as shown in Figure 2. The OFCG is used to generate light wave pulse at 25 GHz with a laser diode windows size of 193.414–193.489 THz, respectively. The end-face of the light wave source is connected to a dual-drive (DD) Mach–Zehnder Modulator (MZM) with three input ports. Among these ports, one is designated for receiving the optical pulse emitted by the LD, while the other two ports are connected to the RF source through electrical amplifiers (EAs), as shown in Figure 2. The output of the DD-MZM is fed to an optical amplifier (OA) followed by the De-Multiplexer (De-MUX) because it contains multiple coherent optical carriers equally spaced from one another. Each port of the OFCG De-MUX arrangement is connected to the corresponding MZM, enabling the upconversion of RF data to the optical domain for individual BBUs in the pool.

Now, the output of each MZM is fed to a MUX arrangement that combines the OFDM-WDM-modulated signal from each BBU and transmits it over the feeder fiber. The feeder fiber is a single-mode fiber (SMF) that spans the optical distribution network (ODN) from the BBUs pool towards the RUs. An optical amplifier (OA) is also employed before the feeder fiber (FF) to compensate for any losses occurring through the OF media and the FSO links.

The long-span feeder fiber is terminated in the remote node (RN) arrangement that employs $1:N$ De-MUX. Here, $N$ represents the number of BBUs and RUs. The End Output ports of the RN De-MUX are connected to their corresponding RUs through three possible channels, including (i) FSO only, (ii) FSO and OF, and (iii) OF only [18,19]. The channel selection depends upon the availability of resources and the need for deployment. However, this work will consider FSO to facilitate a high-capacity and low-latency path between the RN and the corresponding RU, as shown in Figure 2.

The Optical Network Unit (ONU) at the RU module includes a coherent detection module with a local oscillator to convert the data to the RF domain. Subsequently, an OFDM demodulation arrangement is used after the coherent detector to extract the desired information. The down-converted data are then directed to the OFDM demodulation arrangement, linked to a QPSK decoder module.

3. Analysis of the Proposed Architecture

The given architecture is analyzed in a widely recognized simulation software called Optisystem by developing a working model concerning Figure 2 [20]. The simulation setup is shown in Figure 4, and the performance parameters for each simulation model component are outlined in

Table 1. System parameters and corresponding values.

Parameters	Values
Modulation	OFDM QPSK
Data per wavelength	10 Gbps
Subcarriers per BBU	512
Bandwidth of multiplexer	10 GHz
Roll-off factor of LPF	0.2
Feeder fiber length	20 km
Fiber attenuation	0.25 dB/km
Fiber dispersion	18 ps/nm/km
SMF non-linear parameters	Active
FSO transmitter aperture	5 cm
FSO receiver aperture	20 cm

. The simulation setup is developed concerning the system model in Figure 2, with four BBUs and RU modules utilizing four OLTs and ONUs, respectively. Furthermore, the analysis is performed for four transmitter and receiver modules representing four BBUs and RUs, respectively. Each BBU module utilizes 512 subcarriers per band to transmit information to the RU. Moreover, each BBU is assigned a specific wavelength for WDM modulation. Each wavelength is generated through the given OFCG module at 25 GHz in a window of 193.414–193.489 THz, as shown in Figure 5.

The performance of the proposed model is initially analyzed for its capability to transmit high data rates over the FSO links with different channel attenuation. The evaluation of FSO links typically includes the assessment of the turbulence attenuation encountered by the transmitted signal through the open space channel. Turbulence, also known as scintillation, refers to the random fluctuations seen in the intensity of the received signal. These fluctuations are primarily induced by atmospheric non-uniformities, such as variations in pressure and temperature. Numerous models have been suggested to analyze the impact of channel non-uniformities on performance degradation in FSO systems. Some commonly employed models include the Gamma–Gamma, negative exponential, log-normal distribution, K-distribution, etc.

This work utilizes the Gamma–Gamma model for performance analysis. The Gamma–Gamma (GG) model is a versatile approach that can be applied across a wide spectrum of turbulence, spanning from weak to strong turbulence regimes. This model utilizes small $\left(\alpha \right)$ and large $\left(\beta \right)$-scale fluctuators to calculate the normalized light intensity, $I$ [14,21,22]. The probability of $I$, in terms of $\alpha$ and $\beta$, can be written as

$$p\left(I\right)=\frac{2{\left(\alpha \beta \right)}^{\frac{\alpha +\beta }{2}}}{\Gamma \left(\alpha \right)\Gamma \left(\beta \right)}{I}^{\alpha +\frac{\beta }{2}-I}{K}_{\alpha -\beta }(2\sqrt{\alpha \beta I})$$

(1)

$k_{\alpha -\beta }(.)$ in Equation (1) shows the modified Bessel function and $\Gamma \left(.\right)$ is the Gamma function as per the propagation distance d. Now, the values of $\alpha$ and $\beta$, which are the large-scale and small-scale eddies of the scattering process, can be expressed as

$$\alpha ={\left[exp\left(\frac{0.49{\sigma }_l^2}{{\left(1+1.11{\sigma }_l^{2.4}\right)}^{1.17}}\right)-1\right]}^{-1}$$

(2)

$$\beta ={\left[exp\left(\frac{0.519{\sigma }_l^2}{{\left(1+0.69{\sigma }_l^{2.4}\right)}^{0.833}}\right)-1\right]}^{-1}$$

(3)

In Equations (2) and (3), the value of the Rytov Variance $\left({\sigma }_l\right)$ is used to distinguish between the different link atmospheric turbulence regimes. The initial analysis includes the impact of the FSO link on the received power (dBm) and the BER at each RU module over a Gamma–Gamma channel. The index of refraction for the FSO structure is set to ${10}^{-15}{\mathrm{m}}^{-\frac{2}{3}}$ for the analysis in order to incorporate the effect of a moderately turbulent environment. The investigation is conducted with the constraint of achieving an acceptable BER within the $2e^{-3}$ FEC limit [23,24].

Figure 6 shows the BER versus FSO channel length with an attenuation of $1\frac{dB}{km}$, $2\frac{dB}{km}$, and $3\frac{dB}{km}$, respectively. The analysis is performed by implementing the system model in Optisystem, and the BER is measured at the receiving end by employing the BER analyzer. Furthermore, the BER is presented on the logarithmic scale. It is evident from the results that the BER performance degrades as the length of the FSO link between the RN and ONU module increases. This can be attributed to the fact that the BER is proportional to the overall length of the channel because an increase in channel length deteriorates the signal’s power. This degradation in the signal power reduces the signal-to-noise ratio at the receiving end, which results in a higher BER.

For further analysis, it can be observed that the proposed system can transmit more information under less attention from the FSO link. For instance, the best BER values of $7{ \times 10}^{-7}, 1{ \times 10}^{-6},4{ \times 10}^{-6}$ are obtained at the 1 km FSO link for 1 dB/km, 2 dB/km, and 3 dB/km channel attenuation, respectively. Furthermore, performance degradation is observed as channel-induced attenuation becomes more prominent. The lowest performance for all channel attenuations is obtained at a 3 km channel for 3 dB/km attenuation, as it is less than the acceptable threshold.

It can also be observed from Figure 6 that the transmitted signals exhibit varying degrees of attenuation under different weather conditions introduced through channel attenuation. This phenomenon primarily highlights the sensitivity of the system’s performance in the environment present across the FSO link [14]. Additionally, even under the same weather conditions, the transmission distance significantly influences the BER of the signals. For instance, the system provides efficient performance with a BER of $1\times {10}^{-6}$ and $8\times {10}^{-4}$ for 1 and 2 km FSO links at 2 dB/km attenuation. However, the BER becomes equal to $8\times {10}^{-2}$ for the 3 km FSO link at 2 dB/km and crosses the acceptable threshold of $2\times {10}^{-3}$for the nominal performance.

For further analysis, the proposed architecture is evaluated under various turbulent conditions at a link attenuation of ≈3 dB/km by referring to the system performance parameters listed in Table 1. The power at the PIN photodiode is adjusted to identify the optimal conditions for achieving maximum transmission capacity in different turbulent environments, categorized as strong, average, and weak turbulence. Generally, for near-ground and low-altitude free space optical links, the refractive index structure constant $\left({\mathrm{C}}_{\mathrm{n}}^2\right)$ is changed from ${10}^{-13}$ ${\mathrm{m}}^{-\frac{2}{3}}$ to ${10}^{-17}$ ${\mathrm{m}}^{-\frac{2}{3}}$, where values of ${10}^{-13}$ ${\mathrm{m}}^{-\frac{2}{3}}$, ${10}^{-15}$ ${\mathrm{m}}^{-\frac{2}{3}}$, and ${10}^{-17}{\mathrm{m}}^{-\frac{2}{3}}$ represent strong, typical average, and weak turbulence regimes [14]. Furthermore, the received power before a factor of 0.5 dBm varies the coherent detector for each analysis by using an optical attenuator to determine the receiver’s maximum sensitivity.

The BER analysis versus the total received signal power in Figure 7 shows that the receiver sensitivity for channel one in a weak turbulent scenario, for an acceptable BER of $2\times {10}^{-3}$, is approximately −25.3 dBm. Furthermore, an increase in turbulence from ${10}^{-13}$ to ${10}^{-17}$ shows a decrease in receiver sensitivity for channel one from −25.3 to ≈−25 dBm for moderate and ≈−24.8 for high turbulence, respectively, for the same performance parameters.

It can be observed that the FSO channel is more stable under clear environments having less turbulence. It is evident that light signals primarily undergo influences from atmospheric phenomena, such as scattering by suspended particles, molecular absorption, and atmospheric turbulence. However, atmospheric attenuation has a lesser impact on these signals. The scattering effect is directly correlated to the quantity and size of the particles in the atmosphere. More particles result in more severe scattering and, consequently, greater attenuation of the light signals [10,14,15]. However, performance analysis shows that the proposed OFDMA-based WDM-PON, in conjunction with the OFCG at the OLT module and FSO links at the distribution level, can meet the necessary data requirements for next-generation wireless communication networks [2]. This PON architecture is a viable solution enabling seamless communication between the BBU and the corresponding RUs.

For further analysis, the proposed setup is analyzed under different weather conditions to observe system performance when subjected to high atmospheric losses. The analysis chooses three different weather conditions: haze, moderate rain, and heavy rain [25]. The attenuation associated with each weather condition is given in

Table 2. Atmospheric losses.

Weather	Attenuation
Haze	4 dB/km
Moderate Rain	5.8 dB/km
Heavy Rain	9.2 dB/km

. A simulation analysis is made in a strong turbulent environment, and the BER is given in a logarithmic scale by varying the length of the FSO link and the attenuation losses.

It can be observed that the propagation distance decreases with an increase in the overall attenuation across the FSO link, as shown in Figure 8. Furthermore, it can be observed that the lowest performance is obtained for a heavy rain scenario that attenuates the signal at the rate of 9.2 dB/km. As a result of the 9.2 dB/km attenuation in the signal, the proposed model can transmit around a 1.25 km FSO link length within the acceptable BER range. This is evident from the fact that the amount of water droplets increases with an increase in the quantity of rain, as a result of which the level of attenuation increases, thereby increasing the BER [26]. Despite the high attenuation, the proposed model can still support a relatively high transmission range over the FSO link. This can be attributed to the fact that FSO transmission is less affected by rain [25].

On the contrary, the best performance is obtained under hazy weather with an attenuation of 4 dB/km. It can be observed that the proposed system can support data communication until 2 km of FSO link length for an acceptable range of BER. The same can be observed from constellation diagrams that represent a BER of $2\times {10}^{-5}$ for 1 km and $1\times {10}^{-3}$ for a 2 km FSO link. The performance has deteriorated after 2.2 km, and a BER of $4\times {10}^{-2}$ is obtained at 3 km. The same can be observed from the constellation points, which show a significant increase in the noise of the constellation with an increase in the length of the FSO media. This can be attributed to the increase in the signal-to-noise ratio through the influence of the link length and the atmospheric attenuation offered by the channel.

It can be concluded that the performance of the proposed system employing an FSO link at the front-haul section is subject to different performance parameters, like the length of the FSO link, the power of the signal received at the photodiode, the receiver sensitivity, the signal attenuation due to atmospheric losses, and the atmospheric turbulence. An increase in the length of the FSO link aggravates the system’s performance because signal degradation due to atmospheric attenuation and turbulence becomes more prominent, resulting in a decrease in the signal-to-noise ratio that reduces the desirable BER performance.

Energy Analysis

The global carbon footprint has seen a significant increase of 3.5% due to the expansion of the Information and Communication Technologies (ICT) sector in response to customer demand. This rise is primarily attributed to the emission of greenhouse gases (GHG) from communication hardware [27]. The Telecom sector plays a substantial role in this, as it is responsible for about 21% of GHG emissions. Projections indicate that by 2025, its total power consumption will likely reach 0.13 Terawatt (TW), accounting for approximately 7% of the world’s entire electricity supply [28].

Furthermore, the access network is found to account for approximately 70% of the power consumption by the telecommunication networks, significantly increasing its operational expenditure. Although PONs utilize passive components at the ODN, the application of active components at the transmitter and receiver module contributes to approximately 30% and 60% of the total power consumed by the network [29,30]. Consequently, it is of primary importance to determine the proposed model’s energy efficiency to assess the feasibility of deployment for the next-generation communication applications.

This section determines the energy efficiency of the proposed architecture by comparing the total energy consumed by the optical source using FCG against the single light source per OLT in the conventional OFDM PON. The basic architecture of the OFCG employed to generate multiple optical signals is shown in Figure 2, which consists of active and passive components. Because active components are the only source of energy consumption, the total energy consumed by the OFCG can be given by Equation (4):

$$E_{OFCG}=E_{LD}+\left(2\ast E_{EA}\right)+E_{FG}+E_{OA}$$

(4)

where $E_{LD}$ represents the energy consumed by the laser diode installed at the OFCG arrangement, $2\ast E_{EA}$ describes the total energy consumed by two electrical amplifiers, $E_{FG}$ and $E_{OA}$ determine the energy consumed by the Frequency Generator and the optical amplifier, respectively, as shown in Figure 2. The components used at the OLT/BBU side and their energy consumption ratings in Watt (W) are given in

Table 3. Components description and their energy consumption.

Components	Symbols	Energy Consumption (W)
Laser Diode	LD	1.2
OFDM DSP	OFDM	1.2
Analog to Digital	ADC/DAC	1
Optical Amplifier	OA	4.5
Electrical Amplifier	EA	0.5
Frequency Generator	FG	0.5

[31].

Now, the total energy consumed by the OLT at the BBU module in our proposed architecture can be determined by combining the energy consumed by the OFCG in Equation (4) with energies consumed by the OFDM module and DAC, which is expressed in the equation given by

$$E_{T_{P_{OLT}}}=E_{OFCG}+M\ast E_{OFDM}+M\ast E_{DAC}$$

(5)

where M shows the total number of OLTs/BBU and their corresponding ONUs/RU nodes. $E_{OFDM}$ and $E_{DAC}$ define the total energy consumed by the DSP-based OFDM modulator and the digital-to-analog converter module, respectively. Furthermore, $E_{OFCG}$ represents the total amount of energy consumed by the OFCG module.

The energy consumption analysis for the proposed model is performed for 16 OLT modules. The proposed OFCG architecture can generate 16 equally spaced optical pulses used for the modulation process of the RF to the optical signal. For an analysis comparison, the total energy consumed by the conventional OFDM PON, which uses a single laser diode per transmitter and receiver, is given by

$$E_{T_{C_{OLT}}}=M\ast E_{LD}+M\ast E_{OFDM}+M\ast E_{DAC}$$

(6)

here,

$$E_{LD}=M\ast LD$$

(7)

The total energy consumed by the ONU for both the proposed OFDM PON and the conventional OFDM PON architecture is given by

$$E_{T_{P_{ONU}}}=E_{T_{C_{ONU}}}=M\ast E_{CD}+M\ast E_{OFDM-D}+M\ast E_{DAC}$$

(8)

where $E_{CD}$ represents the total energy consumed by the coherent detection module, $E_{OFDM-D}$ and $E_{DAC}$ show the total energy consumed by the OFDM demodulator and the analog-to-digital module, respectively. Because no specific changes are being made at the ONU side of the proposed architecture, the total energy consumed by the proposed and conventional architecture at ONU is considered the same. The total energy consumed by the coherent detection arrangement can be written as

$$E_{CD}=4\ast E_{PD}+2\ast E_{EA}+E_{LD}$$

(9)

Here, Equation (9) shows that four photodiodes are employed at the receiving end to down-convert the signal from the optical to the electrical domain. Moreover, an electrical amplifier and a laser diode are also employed to facilitate the recovery of the intended spectrum.

Table 4. Energy consumption for each component inside the OLT module.

OLTs	${\mathit{E}}_{\mathit{O}\mathit{F}\mathit{C}{\mathit{G}}_{\mathit{P}}}$	${\mathit{E}}_{{\mathit{L}\mathit{D}}_{\mathit{C}}}$	${\mathit{E}}_{\mathit{O}\mathit{F}\mathit{D}{\mathit{M}}_{\mathit{P}}}$	${\mathit{E}}_{\mathit{O}\mathit{F}\mathit{D}{\mathit{M}}_{\mathit{C}}}$	${\mathit{E}}_{\mathit{D}\mathit{A}{\mathit{C}}_{\mathit{P}}}$	${\mathit{E}}_{\mathit{D}\mathit{A}{\mathit{C}}_{\mathit{C}}}$
4	7.2	4.8	4.8	4	4	4
8	7.2	9.6	9.6	8	8	8
16	7.2	19.2	19.2	16	16	16

and Figure 9 demonstrates the total energy consumption for each component inside the OLT against the number of OLTs inside the BBU pool, as per Equation (6) and the system performance parameters in Table 3. It can be observed that most components of the OLT module consume the same amount of energy for both the conventional and proposed architectures. The major difference exists in the light source generation module, where the proposed architecture employs the OFCG and the conventional architecture uses LD for each OLT per BBU. For a limited number of users (four), the conventional architecture consumes less energy owing to the additional number of components in the OFCG module. However, as the number of OLTs increases, the total amount of energy consumed by the OLT module for the proposed architecture starts a downward trend.

This can be explained by the fact that the total cost of energy consumed by the OFCG is determined by Equation (4), implying that energy consumption will decrease as the number of OLTs per BBU increases. A single LD is employed per BBU pool to generate the required spectrum for WDM using the frequency comb generation technique to further explain this phenomenon. This approach allows for the distribution of energy consumption by the LD over the total number of OLTs. Furthermore, the OFDM and associated DAC modules are the only parameters that contribute to the rise in energy cost. However, their contribution towards energy consumption is less dominant than the LD. Therefore, the overall energy consumption cost keeps reducing as we increase the number of OLTs per BBU pool.

For further analysis, $\frac{E_{OL{T}_C}-E_{OL{T}_P}}{E_{OL{T}_C}}\ast 100$ is used to formulate the energy difference between the proposed and conventional architectures. Here, $E_{OL{T}_C}$ represents the total energy consumed by the OLT module for the conventional architecture, whereas $E_{OL{T}_P}$ is the energy for the OLT module of the proposed architecture. It is assumed that the ONU modules in both architectures consume the same amount of energy, as per Equations (8) and (9) and Figure 10. This is because all components utilized for ONU modules in the conventional and proposed architectures are the same. Therefore, ONU modules consume the same amount of energy for both setups, irrespective of the number of OLTs inside the BBU pool.

Table 5. Energy difference between OLT modules of the conventional and proposed architectures.

OLTs	${\mathit{E}}_{{\mathit{T}}_{{\mathit{C}}_{\mathit{O}\mathit{L}\mathit{T}}}}$	${\mathit{E}}_{{\mathit{T}}_{{\mathit{P}}_{\mathit{O}\mathit{L}\mathit{T}}}}$	% Difference
4	13.6	16	−17.6
8	27.2	24.8	8.82
16	54.4	42.4	22.0

shows the energy difference for the OLT modules for both the proposed and conventional architectures. It can be observed that the conventional setup is performing better in terms of energy consumption for a limited number of users. However, as the number of users increases, the proposed architecture with the application of the OFCG module performs efficiently in terms of energy consumption at the BBU pool. The analysis shows that the proposed setup can save 22% of the energy at the BBU pool compared to the conventional architecture.

For further analysis, a comparison of the total energy consumed by the OLT and ONU modules is given for the conventional and proposed architectures as Figure 11. The analysis is made by dividing the total energy consumed by the number of OLTs per BBU pool. It can be observed that the proposed architecture is more efficient in terms of energy consumption for 8, 16, and 32 users. Furthermore, it can be observed that an energy difference of approximately 11% is obtained for the overall system by applying OFCG modules at the BBU pool.

Thus, it can be concluded that the utilization of the OFCG module significantly impacts the consumption of energy inside the BBU pool. The analysis reveals an energy savings of 22% for the proposed OLT compared to the conventional OLT setup. Moreover, employment of the OFDM-WDM-based hybrid PON architecture enables data rate connectivity of up to 10 Gbps per channel over nominal lengths of the FSO link under adverse weather conditions.

4. Conclusions

Recent trends in the growth of internet users have seen exponential growth in terms of mobile subscribers and the amount of data. Consequently, the Telecom sector contributes around 21% of GHG emissions. Its total power consumption is expected to reach a very high level by 2025, making up around 7% of the world’s electricity supply. Therefore, this work develops an appropriate communication system for B5G and 6G networks capable of sustaining high data rates and delivering necessary capacity with decreased energy consumption. The analysis shows that the OFDM-WDM-based PON architecture can support data rates of up to 10 Gbps in highly turbulent environments of the FSO link. Furthermore, utilization of the OFCG module at the BBU pool is observed to reduce energy consumption by approximately 22% compared to the conventional model. Therefore, it can be concluded that the proposed architecture is suitable for the deployment and adaptability of B5G and 6G networks.