A resilient hybrid FSO/SMF ring network with carrier-less nodes for high-speed internet of underwater things

1 Introduction

The increasing demand for high-capacity, low-latency, and reliable communication networks in challenging environments has significantly accelerated research in optical wireless communication technologies [1]. In particular, free space optics (FSO) and underwater wireless optical communication (UWOC) have emerged as promising solutions for applications where conventional radio frequency (RF) and acoustic communication (AC) systems suffer from bandwidth scarcity, high latency, and severe attenuation [1, 2]. These optical technologies offer multi-gigabit data rates, enhanced security, and immunity to electromagnetic interference, making them attractive for next-generation access networks, smart infrastructure, and emerging IoUT applications [3]. However, harsh atmospheric conditions in terrestrial links and turbulence-induced impairments in underwater environments pose major challenges to link reliability, motivating the researchers to explore the possibilities of realizing resilient network architectures capable of sustaining high performance under adverse channel conditions [1, 4]. A promising solution to the connectivity issues in IoUTs is therefore the convergence of UWOC and FSO systems [5]. Despite this potential, the performance of optical signals in FSO and UWOC channels is often hindered by factors like atmospheric attenuation, intensity scintillation, water absorption, and geometric losses [4].

Hybrid optical interconnects integrate different optical communication technologies into a single system to capitalize on their respective strengths, thereby overcoming individual limitations to enhance performance, flexibility, and reliability [6]. Common examples include combinations such as FSO/RF, UWOC/Acoustic, Fiber/FSO, UWOC/Fiber, FSO/UWOC, and UWOC/Fiber/FSO [6]. Among these, the FSO/SMF based hybrid links are particularly notable. By combining free space and fiber optics, it can be a viable solution for reliable, high-speed, and long-distance transmission for IoUT applications [1]. A key advantage of this specific hybrid link is that it avoids the need for WT between the visible and infrared spectra, as often required in other hybrid links involving FSO, SMF, and UWOC technologies while improving communication reliability and extending transmission range [1].

A most relevant application of such resilient optical networks is illustrated in the hybrid terrestrial/underwater optical ring scenario shown in Figure 1.

In this scenario, underwater sensor nodes, autonomous platforms, and monitoring devices are interconnected via UWOC links and seamlessly integrated with terrestrial FSO nodes deployed along coastal regions or offshore platforms. The ring topology enables full-duplex data exchange between underwater nodes and an onshore central office while providing inherent redundancy against node or link failures at the expense of employing the WT relay at the water-air interface. Such an architecture is highly suitable for IoUT-driven applications including coastal surveillance, marine environmental monitoring, offshore energy infrastructure supervision, and disaster early-warning systems where uninterrupted high-speed connectivity between underwater and terrestrial domains is critical.

2 Literature review

Extensive research has been reported on high-speed hybrid optical networks for various applications. A UWOC/SMF/FSO hybrid link employing a photo-detection, remodulate, and forwarding (PRF) relay with intensity modulation/direct detection (IM/DD) was demonstrated for an 8 × 1 Gbps underwater optical wireless sensor network (UOWSN) [1]. An optical/acoustic hybrid link with power control-aided routing protocol was proposed for UOWSNs [3], while a five-wavelength polarization multiplexing scheme enabled a converged FSO/UWOC link [5]. For indoor IoT applications, a visible light communication/optical camera communication (VLC/OCC) hybrid system was developed [6], and a UWOC/SMF hybrid link was designed for monitoring underwater temperature and pressure for sea divers [7]. Multi-hop convergent FSO/UWOC links were investigated for reliable communication between islands [8], and dual-hop FSO/UWOC hybrid links were analyzed for air-to-ocean transmission [9]. Cooperative relaying was explored in VLC/RF hybrid links [10], and FSO/VLC hybrid links utilizing four-level pulse amplitude modulation (PAM4) and non-orthogonal multiple access (NOMA) were reported [11]. A wavelength division multiplexing (WDM) UWOC/FSO hybrid link based on PAM4 was also demonstrated [12]. High altitude platform station (HAPS) integrated FSO/RF/UWOC hybrid links employing decode and forward (DF) relays [13] and unmanned aerial vehicle (UAV) based dual-hop FSO/UOWC hybrid links [14] were investigated for extended coverage. FSO/SMF hybrid links combining WDM, polarization division multiplexing (PDM), and quadrature amplitude modulation (QAM) were developed [15] along with bidirectional integrated SMF/FSO/5G radio links incorporating FBG sensors [16]. Finally, cooperative FSO/acoustic hybrid links supported by fixed-gain amplify-and-forward (AF) relays were analyzed [17]. To further elaborate on the literature review, the major achievements of previously published studies are presented in Table 1 and finally compared with the proposed work.

TABLE 1. Elaboration of the literature survey and comparison with proposed work.

Against this background, the integration of terrestrial FSO and underwater optical nodes into a unified ring topology represents a logical and timely advancement. While some recent studies have hinted at hybrid optical networking concepts, systematic investigations of full-duplex hybrid FSO/SMF ring systems with carrier-less nodes, realistic turbulence modeling, and detailed survivability analysis remain scarce. In particular, the impact of node failures, primary link disruptions, and path inflation in such heterogeneous optical rings has not been sufficiently addressed in the existing literature. In this paper, we address these gaps by proposing and simulating a hybrid FSO/SMF ring network supporting simultaneous DS and US transmissions to multiple terrestrial and underwater nodes without requiring local optical sources and expensive WT scheme. A combined optical signal containing five DPSK-modulated wavelengths each carrying 10 Gbps DS data is sent to the nodes. Each node uses delay-line interferometer to recover the baseband data and reuses the same optical carrier to transmit 10 Gbps US data back to the central office (CO) using OOK modulation. The system enables full-duplex transmission through a specially designed pulse carver based on a DD-MZM and sinusoidal RF source. The performance of the DS and US channels is analyzed using BER results obtained for different values of refractive index structure parameter and atmospheric attenuation coefficient using Gamma-Gamma channel model. Simulation results clearly show that forward-error correction (FEC) target BER of 3.8 × 1 0 − 3 is achieved for DS and US channels under different turbulence strengths and weather conditions.

Based upon above detailed discussion, the main contributions of this study are:

• A novel hybrid FSO/SMF ring network architecture is proposed enabling full-duplex communication with multiple terrestrial and underwater nodes while eliminating the need for local optical sources and WT scheme at the water-air interface.
• The proposed design utilizes DPSK modulation for DS data transmission to improve turbulence resilience and reuses the same optical carriers for US data transmission encoded in OOK modulation format through a DD-MZM based pulse carving scheme.
• System performance is validated through BER analysis under varying atmospheric conditions using Gamma-Gamma channel model, demonstrating FEC compliant transmission with BER of 3.8 × 1 0 − 3.
• Detailed node survivability and link inflation analysis are performed for different scenarios.

The proposed work is simulated using OptiSystem 22 commercial software [18].

3 FSO system and channel model

Line-of-sight (LOS) terrestrial FSO links are established between the central office (CO) and all nodes in the proposed hybrid FSO/SMF ring architecture. The total received optical power for each LOS FSO link can be expressed by Equation 1 [11].The refractive index of the FSO channel frequently fluctuates due to variations in air temperature, humidity, and pressure [4, 11]. The resulting random variation in the intensity of the received signal, known as intensity scintillation or atmospheric turbulence [4, 11], is commonly modeled using the Gamma-Gamma distribution [4, 11]. This channel model is particularly suitable for estimating turbulence-induced fading in FSO systems as it accounts for both small-scale and large-scale fading eddies [19]. Compared to other models, the Gamma-Gamma distribution provides a more accurate estimation of turbulence effects [11]. The probability density function (pdf) of the received signal intensity under Gamma-Gamma channel model is given by Equation 2.Where, K α − β and Γ ( ⋅ ) represent the modified Bessel and Gamma functions, respectively [4, 11]. The Gamma function is defined by Equation 3. α and β parameters can be expressed by Equations 4, 5. σ R 2 can be expressed as [4, 19].In Equation 6, C n 2 parameter quantifies the strength of atmospheric turbulence. In this work, various weather conditions and turbulence regimes are considered, which are characterized by α atm and C n 2 parameters with their respective values listed in Table 2.

P r = P t λ 4 π d 2 G t G r η t η r × 1 L (1)

ρ I = 2 α β α + β 2 Γ α Γ β I α + β / 2 − 1 K α − β 2 α β I (2)

Γ d = ∫ 0 ∞ e x p − t t d − 1 d t (3)

α = e x p 0.49 σ R 2 1 + 1.11 σ R 12 / 5 7 / 6 − 1 − 1 (4)

β = e x p 0.51 σ R 2 1 + 0.69 σ R 12 / 5 5 / 6 − 1 − 1 (5)

σ R 2 = 1.23 C n 2 k 7 / 6 d 11 / 6 , (6)

TABLE 2. Turbulence and attenuation parameters.

4 Working principle of pulse carving scheme

Figure 2 explains the operating principle of proposed pulse carving scheme based on DD-MZM and sinusoidal RF source employed in the ring architecture for CW to pulse signal conversion to implement the transmission of full-duplex data in the ring. The output optical field for an input optical field is expressed by Equation 7 [20, 21].The transfer characteristic of the DD-MZM depicting the output power as a function of applied input voltage is presented in Figure 2a. To transform the CW signal into pulsed form, the DD-MZM is biased at its maximum transmission point and driven by a sinusoidal electrical RF signal as shown in Figure 2b that switches it between its minimum and maximum transmission states. In this work, the sinusoidal drive signal operates at a frequency of 5 GHz, which corresponds to half of the target data rate of each channel (i.e. 10 Gbps). This configuration generates optical pulses with a 33% duty cycle and 10 GHz repetition rate as shown in Figure 2c. The step-by-step mathematical derivation of this pulse carving process is as follows.

E o t = E i t 2 exp j π V 1 t V π + exp j π V 2 t V π (7)

The pulse carving process begins with a continuous-wave (CW) optical signal generated by a laser source. The electric field of this CW laser entering the MZM is expressed by Equation 8 [21].To carve optical pulses from this CW signal, the MZM is driven by a sinusoidal RF signal combined with a DC bias voltage. The total drive voltage applied to the modulator is given by Equation 9 [21].This drive voltage induces a phase difference between the two arms of the MZM, which is the fundamental mechanism through which the modulator converts phase modulation into intensity modulation. The phase difference is given by [21]. Equation 10 can be expanded to separate the DC and AC components.In this representation, ϕ b = π V DC / V π is the constant phase bias determined by the DC voltage while β = π V m / V π is the phase modulation index determined by the amplitude of the RF drive signal. The optical field at the output of the MZM results from the interference of the optical signals propagating through its two arms. For an ideal MZM with balanced splitting and combining, the output field is related to the input field by [21]. Equation 12 shows that the MZM functions as an optical interferometer where the phase difference between the arms determines the output amplitude through a cosine relationship. Substituting the expression for the phase difference yields (Equation 13). Equation 14 shows the instantaneous optical power at the output of the MZM which is the physically observable quantity, is obtained by taking the squared magnitude of the output field.After performing this calculation, Equation 15 gives the fundamental equation describing the pulse carving process [21].This expression reveals that the output power is a time-varying function that follows the square of a cosine with its argument containing both a constant bias term and a sinusoidally varying component. The result is a periodic pulse train carved from the original CW signal. For pulse carving applications requiring high extinction ratio between the pulse peaks and the suppressed background, specific biasing conditions are employed. The MZM is typically biased at the null point as shown by Equation 16, which corresponds to [21].At this bias point, the modulator operates at its most sensitive region providing maximum suppression of the optical carrier between pulses. Additionally, the modulation amplitude is set close to the half-wave voltage to achieve deep modulation [21].Under these conditions as mentioned in Equation 1, the modulator swings between its minimum and maximum transmission states with each cycle of the RF drive signal generating a clean train of optical pulses suitable for subsequent data modulation. The repetition rate of these pulses equals the frequency of the RF drive signal while the pulse width and shape are determined by the specific values of V m and the bias conditions.

E in t = P o e j ω o t (8)

V t = V DC + V m ⁡ cos Ω t (9)

Δ ϕ t = π V t V π (10)

Δ ϕ t = ϕ b + β ⁡ cos Ω t (11)

E out t = E in t cos Δ ϕ t 2 (12)

E out t = P o e j ω o t ⁡ cos ϕ b 2 + β 2 cos Ω t (13)

P out t = | E out t | 2 (14)

P out t = P o ⁡ cos 2 ϕ b 2 + β 2 cos Ω t (15)

ϕ b = π (16)

V m ≈ V π ⇒ β ≈ π (17)

The key symbols and mathematical notations used across the paper are defined in Table 3.

TABLE 3. List of main symbols and notations used in the paper.

5 Proposed hybrid FSO/SMF ring network

The physical layer design of the proposed ring architecture is illustrated in Figure 3. At the central office (CO), five optical DPSK transmitters operating at 10 dBm power each are employed to transmit DS data to five nodes (i.e., four terrestrial nodes and one underwater node comprising a floating platform and underwater electrical sensor). These transmitters operate at wavelengths of λ 1 = 1550 nm, λ 2 = 1550.8 nm, λ 3 = 1551.6 nm, λ 4 = 1552.4 nm, and λ 5 = 1553.2 nm, each carrying DPSK encoded data at a rate of 10 Gbps. The nodes are spaced 400 m apart along the ring. The detailed structure of each optical DPSK transmitter is shown in inset of Figure 3. The optical DPSK signals are combined using an optical power combiner (PC), yielding a composite signal with an aggregated data rate of 50 Gbps. This combined signal is then fed into a pulse carver which converts the CW signals into pulsed optical signals featuring a 33% duty cycle and 10 GHz repetition rate, as detailed in above section. Figure 4 presents the spectral plot of the combined pulsed optical DPSK signal after pulse carving.

The resulting DS data modulated optical pulsed signals for all five nodes are amplified and directed to a transmitter telescope. The optical amplifier (OA) gain is carefully set to ensure that the combined pulsed signal at the telescope input maintains an average power of 20.4 dBm. The combined pulsed signal is transmitted over the first 400 m FSO link. This FSO channel is modeled using the Gamma-Gamma distribution as described above, to account for signal impairments caused by turbulence and attenuation. At Node-1, the combined pulsed signal is received by a receiver telescope. The internal structure of each node is detailed in Figure 5.

As illustrated, the incoming signal first passes through a wavelength division multiplexing (WDM) drop component which selectively drops a specific channel from the WDM signal while allowing the remaining wavelengths to pass through. Accordingly, the DPSK-modulated DS channel at λ 1 = 1550 nm is dropped at Node-1. The corresponding spectral plot of this dropped channel at Node-1 and four remaining channels are shown in Figure 6a and Figure 6b, respectively.

The dropped DPSK-modulated DS channel at λ 1 = 1550 n m is divided using a 95:5 power splitter (PS), as depicted in Figure 5. As illustrated in the system architecture, 5% of the split signal is directed to a DPSK receiver for DS data recovery while the remaining 95% is remodulated with US data encoded in OOK format. The detailed structure of the US OOK transmitter is provided in inset of Figure 3. At the DPSK receiver, whose internal configuration is shown in inset of Figure 3, the DS channel is demodulated using a delay-line interferometer which converts phase variations of the pulsed DPSK signal into corresponding intensity variations. The resulting electrical signal is then low-pass filtered and passed to a BER estimator as depicted in Figure 3. The BER is evaluated by analyzing the statistical characteristics of the received eye diagram. DPSK modulation is employed for DS transmission because it encodes data solely through phase variations while maintaining a constant optical intensity. This constant intensity property allows the same DPSK-modulated optical signal to be reused for US transmission through intensity modulation by US data. As illustrated in the inset of Figure 3, the optical carrier is re-modulated with 10 Gbps electrical US data encoded in OOK format. At Node-1, the OOK-modulated US channel at λ 1 = 1550 nm is recombined with the remaining four DPSK-modulated DS channels using a PC and amplified by an OA, as shown in Figure 3. The gain of the OA is set to achieve an average power of 20.4 dBm at the transmitter telescope input. The combined signal is then transmitted over a 400 m FSO link towards Node-2. Figure 6c presents the corresponding spectral plot. At Node-2, the DPSK-modulated DS channel at λ 2 = 1550.8 nm is demodulated following the same procedure used at Node-1. This channel is then intensity-modulated with 10 Gbps US data. The resulting two OOK-modulated US channels at λ 1 = 1550 n m and λ 2 = 1550.8 n m are recombined with the remaining three DPSK-modulated DS channels at λ 3 = 1551.6 nm, λ 4 = 1552.4 nm, and λ 5 = 1553.2 nm using a PC and amplified to 20.4 dBm before transmission to Node-3 over another 400 m long FSO link. This process repeats at each node. At Node-3, the DS channel at λ 3 = 1551.6 nm is demodulated and then remodulated with US data. The combined signal now containing three OOK-modulated US channels and two DPSK-modulated DS channels are amplified to 20.4 dBm and forwarded to Node-4. At Node-4, the DS channel at λ 4 = 1552.4 nm is processed in the same way, resulting in a combined signal with four OOK-modulated US channels and one remaining DPSK-modulated DS channel at λ 5 = 1553.2 nm, amplified to 20.4 dBm, and transmitted to Node-5. Node-5 comprises a floating platform anchored to an underwater electrical sensor at a depth of 400 m as illustrated in the inset of Figure 3. In practical underwater sensing applications, analog signals generated by the sensor are digitized using an analog-to-digital converter (ADC), producing a binary data stream. This binary signal is then modulated onto an optical carrier for transmission over the communication channel. Following this principle, the DPSK-modulated DS channel at λ 5 = 1553.2 nm is demodulated at node-5 and the same optical carrier is reused to transmit the US data of the underwater sensor (channel-5) encoded in OOK format. The OOK-modulated US channel is then delivered through a 400 m long single-mode fiber (SMF) to the transmitter telescope mounted on the floating platform. The combined optical signal now comprises five OOK-modulated channels at λ 1 = 1550 nm, λ 2 = 1550.8 nm, λ 3 = 1551.6 nm, λ 4 = 1552.4 nm, and λ 5 = 1553.2 nm. This combined signal is then amplified to 20.4 dBm and transmitted back to the CO over the final 400 m FSO link. At the CO, the received composite signal is demultiplexed and photodetected. The resulting electrical signals are low-pass filtered to remove high-frequency harmonics and then passed to BER estimators.

The key parameters used in the numerical simulations are summarized in Table 4.

TABLE 4. Summary of important simulation parameters used in this work.

6 Results and discussion

Following points are assumed to analyze the performance of DS and US channels in proposed FSO/SMF hybrid ring.

• Four nodes are terrestrial while one node comprises a floating platform anchored to an underwater electrical sensor at a depth of 400 m.
• All nodes are distributed in a ring separated by a fixed distance of 400 m.
• LOS connectivity is assumed throughout the ring among all communicating units.
• Geometrical loss is considered in FSO channels while attenuation and dispersion of standard SMF is assumed.
• Thermal noise of photodetector in Node-5 is ignored due to low water temperature.
• FEC target BER of 3.8 × 1 0 − 3 is considered.
• DS channel-2 and 4 while US channel-1, 3, and 5 are randomly chosen to discuss the BER performance.

To evaluate the performance of DS and US channels in the proposed ring architecture, we have considered different turbulence regimes and weather conditions as characterized by different values of C n 2 and α atm parameters in OptiSystem. BER plots are obtained in line with the general practice that is being followed in literature that is similar to our study. To further clarify the methodology for the benefit of readers, we would like to explain that the BER-versus-received-power curves were generated by varying the received optical power using an optical attenuator placed immediately before the PIN photodetector in the OptiSystem simulation setup. For each specific turbulence regime and weather condition, the attenuator was adjusted to obtain different received power levels and the corresponding BER was recorded at each level. These data points were then used to construct the BER curves. The receiver sensitivity is the minimum received optical power required to achieve the forward-error correction (FEC) target BER of 3.8 × 1 0 − 3.

Figure 7 shows the BER versus received power plots of DS and US channels for C n 2 parameter values of 5 × 1 0 − 16 m − 2 / 3, 5 × 1 0 − 14 m − 2 / 3, and 5 × 1 0 − 12 m − 2 / 3. It may be noticed from Figure 7a that receiver sensitivities of DS channel-2 and 4 are around −25.2 dBm and −24.9 dBm, respectively for C n 2 parameter value of 5 × 1 0 − 16 m − 2 / 3, −24.7 dBm and −24.5 dBm, respectively for C n 2 parameter value of 5 × 1 0 − 14 m − 2 / 3, and −24 dBm and −23.5 dBm, respectively for C n 2 parameter value of 5 × 1 0 − 12 m − 2 / 3 at the FEC target BER. Similarly, it may be observed from Figure 7b that receiver sensitivities of US channel-1, 3, and 5 are around −25.8 dBm, −26 dBm, and −26.6 dBm, respectively for C n 2 parameter value of 5 × 1 0 − 16 m − 2 / 3, −24.4 dBm, −25 dBm, and −25.5 dBm, respectively for C n 2 parameter value of 5 × 1 0 − 14 m − 2 / 3, and −15.7 dBm, −16.5 dBm, and −20.6 dBm, respectively for C n 2 parameter value of 5 × 1 0 − 12 m − 2 / 3 at the FEC target BER. The system experiences a progressive degradation in perforamnce of all channels with increasing turbulence strength. The reason of observed trend in receiver sensitivities of DS and US channels for all three values of C n 2 parameter can be understood considering Equation 6. It may be observed from Equation 6 that the Rytov variance depends upon the range of the FSO channel and turbulence. Therefore, a channel transmitted over longer FSO link has poor performance compared to the channel propagating over a shorter FSO link. Figure 3 shows that the DS channel-2 covers the transmission distance of 800 m compared to DS channel-4 which covers the distance of 1,600 m. Similarly, US channel-1 covers the longest transmission distance of 2,400 m, US channel-3 covers the transmission distance of 1,200 m, and US channel-5 covers the shortest distance of 400 m over FSO link. Therefore, the channel sensitivities are representative of the distance covered over FSO link both for DS and US channels. It is also evident from Figure 7a and Figure 7b that overall penalties of 1.2 dB and 1.4 dB are observed in received optical power for DS channel-2 and 4 for C n 2 parameter values of 5 × 1 0 − 16 m − 2 / 3 and 5 × 1 0 − 12 m − 2 / 3, respectively. Similarly, overall penalties of 10.1 dB, 9.5 dB, and 6 dB are observed in received optical power for US channel-1, 3, and 5 for C n 2 parameter values of 5 × 1 0 − 16 m − 2 / 3 and 5 × 1 0 − 12 m − 2 / 3, respectively. To further evaluate the performance of the DS and US channels, the eye diagrams of DS channel-1, 3, and 5 are obtained for different values of C n 2 parameter, as shown in Figure 8.

It can be observed that as the turbulence increases, the eye diagrams of all DS channels progressively degrade and narrow due to increased intensity fluctuations. Moreover, because channel-5 traverses a longer FSO path, its eye diagrams are more severely degraded and exhibit reduced opening compared to channel-1. Nevertheless, the eye openings remain sufficiently wide to enable reliable discrimination between ’0′ and ’1′ at the receiver. Similarly, the eye diagrams of US channel-2 and 4 are obtained for different values of C n 2 parameter, as shown in Figure 9.

It can be observed that as the turbulence increases, the eye diagrams of all US channels progressively degrade and narrow due to increased intensity fluctuations. Moreover, because channel-2 traverses a longer FSO path, its eye diagrams are more severely degraded and exhibit reduced opening compared to channel-4. Nevertheless, the eye openings remain sufficiently wide to enable reliable discrimination between ’0′ and ’1′ at the receiver.

Figure 10 shows the BER versus received power plots of DS and US channels for α atm parameter values of 2 dB/km, 10 dB/km, and 20 dB/km. It may be noticed from Figure 7a that receiver sensitivities of DS channel-2 and 4 are around −25.2 dBm and −24.9 dBm, respectively for α atm parameter value of 2 dB/km, −21.8 dBm and −21 dBm, respectively for α atm parameter value of 10 dB/km, and −20 dBm and −19.8 dBm, respectively for α a t m parameter value of 20 dB/km at the FEC target BER. Similarly, it may be observed from Figure 10b that receiver sensitivities of US channel-1, 3, and 5 are around −25.8 dBm, −26 dBm, and −26.6 dBm, respectively for α atm parameter value of 2 dB/km, −21.9 dBm, −22.5 dBm, and −23 dBm, respectively for α atm parameter value of 10 dB/km, and −20.3 dBm, −20.5 dBm, and −21 dBm, respectively for α atm parameter value of 20 dB/km at the FEC target BER. The system again undergoes a progressive degradation in perforamnce of all channels with increasing the value of atmospheric attenuation coefficient. The observed trend in receiver sensitivities for both DS and US channels, where performance degrades progressively with higher atmospheric attenuation values, can be attributed to the fundamental principles of optical signal propagation through the FSO channel. The atmospheric attenuation coefficient together with turbulence directly quantifies the power loss per unit distance that optical signals experience as they travel through the air. Mathematically, it can be understood considering Equation 1. When this coefficient increases from 2 dB/km to 20 dB/km, the optical signals encounter substantially higher absorption and scattering losses over the same transmission distance. This increased attenuation results in a lower received optical power at the photodetector for any given transmitted power level. To maintain the same BER performance, particularly at the FEC threshold, the system requires higher received power levels, which manifests as poorer receiver sensitivity values. Figure 10 shows that the DS channel-2 covers the transmission distance of 800 m compared to DS channel-4 which covers the distance of 1,600 m. Similarly, US channel-1 covers the longest transmission distance of 2,400 m, US channel-3 covers the transmission distance of 1,200 m, and US channel-5 covers the shortest distance of 400 m over FSO link. Therefore, the channel sensitivities are representative of the distance covered over FSO link both for DS and US channels. It is also evident from Figure 10a and Figure 10b that overall penalties of 5.2 dB and 5.1 dB are observed in received optical power for DS channel-2 and 4 for α atm parameter values of 2 dB/km and 20 dB/km, respectively. Similarly, overall penalties of 5.5 dB, 5.5 dB, and 5.6 dB are observed in received optical power for US channel-1, 3, and 5 for α atm parameter values of 2 dB/km and 20 dB/km, respectively.

To further evaluate the performance of the DS and US channels, the eye diagrams of DS channel-1, 3, and 5 are obtained for different values of α atm parameter, as shown in Figure 11.

It can be observed that as the atmospheric attenuation increases, the eye diagrams of all DS channels progressively degrade and narrow due to increased intensity fluctuations. Moreover, because channel-5 traverses a longer FSO path, its eye diagrams are more severely degraded and exhibit reduced opening compared to channel-1. Nevertheless, the eye openings remain sufficiently wide to enable reliable discrimination between ’0′ and ’1′ at the receiver. Similarly, the eye diagrams of US channel-2 and 4 are obtained for different values of α atm parameter, as shown in Figure 12.

It can be observed that as the atmospheric attenuation increases, the eye diagrams of all US channels progressively degrade and narrow due to increased intensity fluctuations. Moreover, because channel-2 traverses a longer FSO path, its eye diagrams are more severely degraded and exhibit reduced opening compared to channel-4. Nevertheless, the eye openings remain sufficiently wide to enable reliable discrimination between ’0′ and ’1′ at the receiver.

To show the effect of turbulence and atmospheric attenuation on Q factor, Figure 13 shows the Q factor versus received power plots of DS channel-2 and US channel-5 for different values of C n 2 and α atm parameters. Figure 13a and Figure 13b clearly illustrate that the Q factor values for DS channel-2 and US channel-5 improve with higher received optical power at photo-detectors. This trend may be attributed to a fundamental trade-off that achieving a high Q factor, requires increased optical power at the photo-detector which consequently limits the FSO range due to high turbulence and atmospheric attenuation.

To further demonstrate the effectiveness of the proposed hybrid FSO/SMF ring architecture, we compare its performance with existing state-of-the-art studies reported in the literature. This comparative analysis, summarized in Table 1, considers key performance metrics such as the number of channels, aggregate data rate, total link range, and modulation formats employed in each study. As evident from Table 1, the proposed architecture achieves an aggregate data rate of 100 Gbps for 8 channels which is significantly higher than many previously reported hybrid optical networks. For example, the UWOC/SMF/FSO link demonstrated in [1] achieved only 8 Gbps over 5.04 km with 8 channels using NRZ-OOK modulation, while the proposed system delivers 100 Gbps over a 3.2 km range with the same number of channels. Similarly, the FSO/UWOC link reported in [5] achieved 500 Gbps over a relatively short range of 0.215 km using 10 channels with PAM4 modulation, while our architecture provides 100 Gbps over a considerably longer range of 3.2 km, making it more suitable for practical IoUT deployment scenarios. Moreover, compared to FSO/SMF hybrid links reported in [15, 16], which achieved 448 Gbps over 53.5 km and 180 Gbps over 34 km, respectively, our proposed architecture offers a more balanced trade-off between data rate, link range, and channel count. While those studies employed advanced modulation formats such as QAM and OFDM, our system employs simpler DPSK/OOK modulation, which reduces system complexity and cost while still delivering competitive performance. In addition, our architecture eliminates the need for WT at the water-air interface, which is a major challenge in traditional FSO/UWOC systems. This simplification reduces system complexity, improves reliability, and enhances overall network scalability. The inclusion of the pulse carving scheme for full-duplex operation further distinguishes our work from existing studies, as it enables efficient bidirectional data transmission within the ring architecture. Overall, the comparative analysis confirms that the proposed FSO/SMF ring architecture offers a compelling solution for high-speed IoUTs applications, achieving a favorable combination of data rate, link range, number of channels, and system simplicity compared to existing state-of-the-art approaches.

7 Node survivability and link inflation analysis

7.1 Node survivability analysis

The reliability of each node is determined by its constituent components, as shown in Figure 5. Each node comprises a receiver telescope, WDM drop component, 95:5 power splitter, DPSK receiver, OOK transmitter, power combiner, optical amplifier, and transmitter telescope. Assuming constant failure rates, the reliability of a node at time t is given by:In Equation 18, λ n = ∑ j = 1 m λ j is the sum of failure rates of all m components within a node. For the end-to-end connection from the CO to Node k, the signal must traverse all preceding links and the target node itself. The reliability of this path is given in mathematical form by Equation 19.Where, R l , i ( t ) = exp − λ l L i t is the reliability of the FSO link between Node i − 1 and Node i, with length L i = 400 m and failure rate per kilometer λ l. Node-5 includes an additional 400 m SMF segment connecting the floating platform to the underwater sensor. Its effective reliability is.In Equation 20, λ SMF is the failure rate of SMF per kilometer. Network survivability under single node failure is defined as the probability that at most one node fails assuming identical and independent node failures, is given by Equation 21 [22].For N = 5 nodes, this simplifies to Equation 22.The probability that all FSO links remain operational is expressed by Equation 23.With total FSO link length of 2,400 m (six links of 400 m each), this becomes R links ( t ) = exp − 2.4 λ l t. The SMF link at Node-5 has reliability R SMF ( t ) = exp − 0.4 λ SMF t. The overall network reliability, considering node failures, all FSO link failures, and the SMF link failure is given by Equation 24:For numerical illustration, assuming λ n = 1 0 − 5 failures/hour, λ l = 1 0 − 6 failures/hour/km, and λ SMF = 1 0 − 6 failures/hour/km [23, 24], the node reliability at t = 8760 hours (1 year) is R n ≈ 0.916. The network survivability under node failures alone is S node ≈ 0.94, indicating a 94% probability that at most one node fails within the first year. The probability that all FSO links remain operational is R links = exp ( − 2.4 × 1 0 − 6 × 8760 ) = exp ( − 0.021024 ) ≈ 0.9792. The SMF link reliability is R SMF = exp ( − 0.4 × 1 0 − 6 × 8760 ) = exp ( − 0.003504 ) ≈ 0.9965. The overall network reliability is S network ≈ 0.94 × 0.9792 × 0.9965 ≈ 0.9172, or 91.72%.

R n t = exp − λ n t (18)

R C O → k t = ∏ i = 1 k − 1 R l , i t × R n t (19)

R n 5 t = R n t × exp − λ SMF × 0.4 × t (20)

S node t = N 1 1 − R n R n N − 1 + R n N (21)

S node t = R n 4 5 − 4 R n (22)

R links t = ∏ i = 1 6 exp − λ l L i t = exp − λ l t ∑ i = 1 6 L i (23)

S network t = R n 4 5 − 4 R n × exp − 2.4 λ l t × exp − 0.4 λ SMF t (24)

7.2 Link inflation analysis

Path or link inflation refers to network traffic taking a longer route than necessary often because of routing policies, inefficient topology, or peering, leading to increased latency [25]. Mathematically, link inflation is the ratio of the protection path length to the primary path length when a failure occurs [25]. In the proposed ring architecture (between points “A″ and “B″), only Node-1 is provided with a dedicated secondary FSO link parallel to the primary FSO link between the CO and Node-1 as shown in Figure 14 and it represents a corporate body requiring uninterrupted connectivity.

All other nodes have no protection and rely solely on the primary ring path. This analysis considers the failure of the primary CO–Node-1 link and the subsequent switch to the secondary path. For Node-1, the primary DS path length from CO is P D S ( 1 ) = 0.4 km. When the primary CO–Node-1 FSO link fails, DS traffic is automatically rerouted through the dedicated secondary FSO link which is identical in length to the primary link. Therefore, the protection path length for DS transmission remains L prot,DS ( 1 ) = 400 m. The link inflation factor for DS transmission to Node-1 is computed as given in Equation 25:This indicates that there is no path inflation for Node-1 when the secondary path is utilized as the protection link is identical in length to the primary link. For Node-2 through Node-5, no protection paths exist. If any link along their route fails, connectivity is lost entirely. Therefore, link inflation is not applicable for these nodes under the current protection scheme. Node-5 includes an additional 400 m SMF segment connecting the floating platform to the underwater sensor. Under normal operation, the DS path length for Node-5 is P D S ( 5 ) = ( 5 × 400 ) + 400 = 2400 m. Under the CO–Node-1 link failure, the DS path length for Node-5 remains unchanged as the secondary link simply replaces the failed primary link with an identical length link, and the SMF segment is unaffected. Therefore, no inflation occurs for Node-5 either.

I D S 1 = L prot,DS 1 P D S 1 = 0.4 0.4 = 1 (25)

8 Conclusion

A hybrid FSO/SMF ring network supporting simultaneous downstream and upstream transmissions to multiple terrestrial and underwater nodes has been successfully proposed and simulated. The architecture eliminates the need for local optical sources and expensive wavelength translation at the water-air interface by employing DPSK modulation for downstream data transmission, which enhances turbulence resilience, and reusing the same optical carriers for upstream OOK modulation through a specially designed DD-MZM based pulse carving scheme. The system performance has been rigorously evaluated under varying atmospheric conditions using the Gamma-Gamma channel model, with simulation results confirming that the forward-error correction target BER of 3.8 × 1 0 − 3 is achieved for both downstream and upstream channels across different turbulence strengths and weather conditions. Additionally, node survivability analysis demonstrates robust network resilience under single node failure scenarios, while link inflation analysis for the protected Node-1 confirms zero path inflation due to the dedicated secondary FSO link identical in length to the primary path. These findings collectively validate that the proposed ring topology offers a flexible, cost-effective, and resilient solution for high-speed, long-range future Internet of Underwater Things (IoUTs) applications, addressing critical gaps in existing hybrid optical communication systems.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. To generate images in this work.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Abhishek Sharma, National Institute of Technology, Hamirpur, India

Reviewed by: Xiaoqing Tian, Hangzhou Dianzi University, China

Kanika Chauhan, Indira Gandhi Delhi Technical University for Women, India