1 Introduction

Internet traffic is growing exponentially day by day, so we need faster transmission system which can be achieved with faster photonic devices such as AOWC, all-optical logic gates, optical amplifier and optical switches [1–2]. These photonic devices are based on nonlinear applications such as XGM, four-wave mixing (FWM) and cross phase modulation (XPM) of SOA. There are two prominent ways to boost WDM networks capacity, in which the first way is by increasing the density of independent channels over an existing fiber and the second way is to enable the spatial reuse of the wavelengths exploiting the wavelength conversion. In order to optimize the spectral efficiency, switching, and transmission of the new generation data center networks (DCNs), dense wavelength division multiplexing (DWDM) is a commonly used technology. Additionally, high-speed optical wavelength conversion is being proposed prominently for robust and capable intra-DCN switching [3–4].

Previous proposals have included a wide range of ideas, exploiting nonlinearities such as four-wave mixing, highly nonlinear fibers (HNLFs), Erbium-doped fiber laser, and semiconductor optical amplifiers (SOAs) [5,6,7,8,9,10]. A spectrally efficient modulation technique is required for transmission of signals for wavelength conversion [11]. T. Richter et al. [12] have investigated wavelength conversion and parametric amplification of 112 Gbit/s DQPSK signals in both polarization-dependent (straight) as well as in a polarization-independent (loop) configuration. They have exploited dual-pump four-wave mixing (FWM) in HNLF to achieve 20 dB ON-OFF gain using single polarization and dual polarization DQPSK to achieve 0.4 dB OSNR penalty. Utilizing phase conjugation in highly nonlinear fiber (HNLF), polarization independent parametric amplification and wavelength conversion of a 2.048-Tbit/s 8-WDM PDM 16-QAM signal have been conceived [13]. In order to achieve the 10 dB ON-OFF gain, 9 dB conversion efficiency (CE), and broadband flat gain for converted wavelength, they have taken use of the counter phase modulation (CPM) in the fiber optical parametric amplifier (FOPA) incorporating twin pumps.

J. Lu et al. [14] have demonstrated the polarization insensitive all optical wavelength conversion (AOWC) based on dual-pump four-wave mixing (FWM) in a SOA employing the polarization multiplexing. They have used both parallel pumps and orthogonal pumps and observed that translated signal at different wavelength can be isolated into two lightwaves without any cross-talk. Moreover, they have concluded that SOA injection current, modulation data rates and frequency spacing play a major role in the performance of the system. At a line rate of 40 GBaud, R. Elschner et al. [15] have studied the parametric wavelength conversion of DPSK and DQPSK signals by taking use of the pump phase modulation. They have employed the four-wave mixing in highly nonlinear fibers and the tunable fiber Bragg grating (FBG) filter. D. Z. Hsu et al. [16] have demonstrated that an assist beam can improve the SBR of wide-band SOA wavelength converters and the conversion efficiency. They have exploited an intense light beam of short wavelength to a twin-pumped SOA to optimize the CE and ratio of signal-to-noise (SNR) by up to 8.5 and 4.1 dB respectively. Furthermore, they have determined that conversion ratio can be more than 10 dB over a span of 80 nm wavelength by utilizing the SOA with 23 dB gain, and that the power loss factors for upside and downside frequency shift at 10 Gb/s transmissions are less than 0.5 dB and 2.5 dB, respectively. Using the FWM in SOA, Benoît Filion et al. [17] have imagined the wavelength conversion of 16 GBaud 16-QAM and 5 GBaud 64-QAM signals over the whole wavelength range of the C-band. Using co-polarized pumps, they have measured the bit-error rate below the FEC threshold. The SOA-based FWM wavelength converter’s performance at 2.5 Gbps has been demonstrated by F. D. Mahad et al. [18]. They have noted the range of frequency shift, optical SNR, and CE for both higher side and lower side translated idlers. They have also detected a notable increase in OSNR. Since conversion efficiency decreases as input power to SOA grows, a trade-off between OSNR and SOA input power is required. Double-Stage Cross-Gain Modulation in SOAs has been exploited for WDM multicasting using intrinsically stable and polarization independent circuit structure [19]. Simultaneous multiwavelength conversion of an input signal of 10 Gbps to eight 200-GHz-spaced output channels has been reported based on the principle of synchronization between an incoming data-modulated signal and various local continuous wave lasers. This format consists of two synchronous cascading XGM processes and its simple configuration provides intrinsic stability. Polarization maintenance reduces the interchannel crosstalk to realize the less power penalty of 2.5 dB. X. Kong et al. [20] have investigated AOWC based on FWM nonlinearity using dual users coherent optical orthogonal frequency division multiplexing employing cascaded HNLFs. Optical spectra and constellation diagrams verify the results. The system achieves better phase noise resistance and error tolerance compared to one without wavelength conversion. At 60 Gbps, high-order 64-QAM signals have undergone wavelength conversion [21]. By utilizing the FWM in HNLF at 1551.38 nm signal wavelength, idler constellation pattern analysis has been used to optimize operational parameters and minimize the power penalty up to 0.3 dB. The notion of polarization-independent wavelength conversion in the C-band spectrum via phase modulation (PM) in highly nonlinear impaired fiber was developed by S. Takasaka et al. [22]. Using polarization scrambled 10Gbit/s NRZ and DPSK signals, they have achieved less than 0.3 dB power penalty in the whole C band window. J. Mirza et al. [23] have demonstrated a coherent tunable interference free microwave photonic notch filter with two negative and positive coefficients produced by using polarization multiplexing and cross gain modulation in a SOA. Two orthogonal polarization states of a single continuous wave laser diode are used to realize the negative and positive coefficients of the notch filter. The differential group delay of a birefringent single-mode fiber is adjusted to adjust the free spectral range of a microwave photonic filter. Impact of cross gain modulation is realized and good tunability is observed.

F. Bontempi et al. [24] have demonstrated 10 Gbps novel AOWC using an Indium Phosphide monolithically integrated optical circuit for both unicast (single output wavelength) and multicast (many output wavelengths) conversion. They have taken advantage of 10 Gb/s NRZ-OOK signals in SOA using two stages of cascaded XGM.They have experienced mild power penalty of 2 dB/2.6 dB inherent in the bit error rate for unicast/multicast scheme, respectively using photonic integrated circuit using optical gating technique. Numerous more AOWC techniques, including format-free wavelength conversion, coherent type conversion, and optical gate switch type, have been envisaged in the past. Slow gain recovery for high speed AOWC is a drawback of SOA-based optical processing; to address this problem, SOA is followed by a detuned frequency selector block, which significantly lowers OSNR [25]. Concurrent reshaping using cross-gain compression resulted from cascaded SOAs is another method to address this issue. Another solution to the issue is a turbo switch arrangement that comprised of two stages of cascading SOAs with an optical BPF in between. XGM in the first SOA is exploited by the turbo switch provided with a modulated pump signal and a continuous wave probe signal. Subsequently, the filter exhausts the pump signal prior to it reaching the second SOA. By utilizing the gain dynamics of the second SOA to offset the slow rate of the first SOA gain response, this architecture produces extremely quick gain recovery rates. Turbo switch reduces the need for differential mechanisms.

M. Smit et al. [26] have studied advanced usage of PIC through Multi Project Wafer services on a single chip to reduce the cost within reach for many SMEs. This technology can be expanded in various other CMOS technology to interact with electronic functionality. Monolithic photonic integrated circuit turbo-switch realization and characterization is the new dimensions to explore the potential usage of faster optical signal processing in DWDM networks. Recently, some PIC configurations have been realized and characterized using different modulation and filtering combination but slow data rate and conversion efficiency are major concerns to deal with. Furthermore, data rates and power penalties can be optimized further using polarization multiplexed emulator and spectrally efficient modulating configurations. Photonic crystal resonator enables highly precise adjustment of the wavelengths involved in nonlinear conversion processes. Furthermore, the photonic devices offer improved performance and versatility in frequency conversion, wavelength shifting, and signal processing, which holds promise for advancements in a number of domains, including sensing, optical computing, and telecommunications.

In this work, we demonstrate in detail the characterization and performance of InP Photonic Integrated AOWC of 128 GBaud dual polarized 16QAM signals using XGM in SOAs. Indium Phosphide (InP) is the most advanced platform for the photonic integrated circuits (PICs) of active elements such as semiconductor optical amplifiers (SOAs), lasers, modulators and photodetectors. InP PICs reduce the cost, size, weight and power significantly for free space optical communications as it integrates lasers, modulators, amplifiers and detectors in a single platform. The PIC incorporates the two cascaded SOAs to exploit double stage cross-gain-modulation in SOAs with a delay-line interferometer (DLI) filter and band pass filter in between. Figure 1 shows the schematic of the Indium Phosphide -photonic integrated all-optical wavelength conversion of 128 GBaud DP-16QAM signals using XGM in SOAs.

The internal structure of the Indium Phosphide PIC is depicted in Fig. 2, and Fig. 3 displays the experimental setup for the work that is being described. In addition, this work is divided into the following four sections: Sect. 1 provides an overview of current research on wavelength conversion, while the Sect. 2 delves into the theoretical examination of DP-16QAM transmission and XGM within SOAs. The device’s construction and functioning are briefly discussed in Sect. 3. At last, the findings have been examined and evaluated in Sect. 4.

Fig. 1
figure 1

Fundamental configuration of the proposed circuit. DP-16QAM: dual-polarization 16 quadrature amplitude modulation, PolMux Emulator: emulator with polarization multiplexing, DSP: digital signal processing, InP-PIC: Indium Phosphide photonic integrated circuit

Full size image
Fig. 2
figure 2

Picture of the Indium Phosphide photonic integrated circuit (InP-PIC). DLI: delay-line interferometer, MMI: multi-mode interference

Full size image

2 Theoretical Analysis

All optical wavelength conversion (AOWC) is a device which is used to transfer the data from one wavelength to another without accessing electrical domain. Indium Phosphide photonic integrated circuit (InP-PIC) is employed in proposed experimental setup as shown in Fig. 3. InP-PIC consists dual cascaded SOAs, a delay-line interferometer (DLI) filter and band pass filter as shown in Fig. 2. The principle of cross gain modulation (XGM) is the modulation in the SOA gain due to a strong gain saturation driven by the intensity modulated pump signal. If gain of the employed SOAs is polarization independent then such XGM SOA converters are called polarization insensitive. The XGM is the consequence of reduction in optical gain at a wavelength when the gain is saturated by the optical signal with different wavelength. The principle setup for AOWC is shown in Fig. 1. The DP-16QAM modulation used in this work to generate 128 GBaud optical signal by modulating a CW laser signal with central wavelength \( {\lambda }_{s}\). A lightwave is said to be an ideal carrier for optical transmission if it keeps amplitude, frequency and phase unchanged. However, in practice, there is no perfect carrier but optical light sources have been optimized surprisingly in the last decades. The normalized electric field of an ideal lightwave produced by continuous wave (CW) laser can be expressed as in Eqs. (1),

$$ \overrightarrow{E\left(t\right)}=\sqrt{{E}_{o}}\text{e}\text{x}\text{p}\left(j\left({\omega }_{s}t+{\phi }_{s}\right)\right)\overrightarrow{{e}_{s}}$$
(1)

where, the character “s” indicates laser signal, the \( \sqrt{{E}_{o}}\) represents the field amplitude. \( {\omega }_{s}\) denotes the frequency and \( \overrightarrow{{e}_{s}}\) is the polarization of the lightwave. Equations (2) and (3) show the relationship between bit rate, bandwidth, and rate of symbol transmission (Baud rate) in a multilevel advanced digital modulation.

$$ {\text{S}}_{\text{r}}=\frac{{\text{b}}_{r}}{{\text{l}\text{o}\text{g}}_{2}\left(\text{M}\right)} $$
(2)

and,

$$ \text{B}\text{W}=\frac{{\text{b}}_{r}\left(1+{\upalpha }\right)}{{\text{l}\text{o}\text{g}}_{2}\left(\text{M}\right)}$$
(3)

where, BW stands for bandwidth (GHz), \( {\text{S}}_{\text{r}} \)for representing symbol rate (GBaud), M for the number of symbols dependent on \( {\text{l}\text{o}\text{g}}_{2}\left(\text{M}\right)\) data bits per symbol, and α for the roll-off ratio. Bit rate (Gbps) is represented by \( {\text{b}}_{r}\). \( M=4\) For optical modulation using 16-QAM. The optical IQ modulator (IQM) is exploited within the star 16QAM transmitter. To generate higher-order optical Square QAM constellations multilevel electrical driving signals are required to 16-QAM. Quaternary driving signals, such as quadrature phase and in-phase driving signals for both polarization (X and Y), are needed for dual polarization (X and Y) 16QAM. Equation (4) describes the optical output field of the IQM for QAM using NRZ pulse form.

$$ \overrightarrow{E\left(t\right)}=\sqrt{{E}_{o}}\text{e}\text{x}\text{p}\left(j\left({\omega }_{s}t+{\phi }_{s}\right)\right){A}_{IQM}\left(t\right){\phi }_{IQM}\left(t\right)$$
(4)

In Eq. (4), \( {A}_{IQM}\left(t\right)\) is amplitude modulation of IQM and \( {\phi }_{IQM}\left(t\right)\) is the phase modulation. To accomplish the generation of the in-phase and quadrature driving signals, we require level generators and appropriate encoder, probably by utilizing digital signal processing.

A continuous stream of light wave at the predefined wavelength undergoes the modulation by varying the gain. As it traverses through the SOA, it retains the modulated signal. XGM SOA converters are classified into co-propagation and counter-propagation schemes. XGM SOA converters are classified as co-propagation and counter propagation scheme. XGM schemes generates an idler that is inverse of the input signal. Converter performance is significantly affected by the parameters of SOAs. Converted signal attains higher extinction ratio in longer SOA. The DP-16QAM modulated intense laser signal \( {\lambda }_{s}\) and the continuous wave (CW) probe signal \( {\lambda }_{c}\) are simultaneously coupled into the two stages using 3 dB coupler. For multi-stage wavelength converter, the large signal nonlinear transmission functions are expressed as [27–28] in Eqs. (5) and (6),

$$ \left\{{\left(\frac{{\text{g}}_{1}^{\text{m}}}{\stackrel{-}{{\text{g}}_{1}^{\text{m}}}}\right)}^{\left(\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{${\upgamma }$}\right.\right)}\stackrel{-}{{\text{g}}_{2}^{\text{m}}}-1\right\}{\upgamma }{\mathcal{x}}_{20}^{\text{m}}=\text{l}\text{n}\left(\frac{\stackrel{-}{{\text{g}}_{1}^{\text{m}}}}{{\text{g}}_{1}^{\text{m}}}\right)-{\mathcal{x}}_{10}^{\text{m}}\left({\text{g}}_{1}^{\text{m}}-1\right)$$
(5)

and,

$$ \left\{{\left(\frac{{\text{g}}_{2}^{\text{m}}}{\stackrel{-}{{\text{g}}_{2}^{\text{m}}}}\right)}^{\gamma }\stackrel{-}{{\text{g}}_{1}^{\text{m}}}-1\right\}\frac{{\mathcal{x}}_{10}^{\text{m}}}{{\upgamma }}=\text{l}\text{n}\left(\frac{\stackrel{-}{{\text{g}}_{2}^{\text{m}}}}{{\text{g}}_{2}^{\text{m}}}\right)-{\mathcal{x}}_{20}^{\text{m}}\left({\text{g}}_{2}^{\text{m}}-1\right)$$
(6)

where, \( {\text{g}}_{\mathcal{j}}^{\text{m}} \)represents single pass gain of the mth stage \( (m=1, 2)\) at \( {{\uplambda }}_{\mathcal{j}}, \mathcal{j}=1, 2\). Normalize input power at the mth stage is given as Eqs. (7),

$$ {\mathcal{x}}_{\mathcal{j}0}^{\text{m}}=\raisebox{1ex}{${p}_{\mathcal{j}0}^{m}$}\!\left/ \!\raisebox{-1ex}{${p}_{\mathcal{j} }^{sat}$}\right.$$
(7)

where, \( {p}_{\mathcal{j} }^{sat}=\frac{Ah\nu }{\left({{\uplambda }}_{\mathcal{j}}?{\tau }_{sc}{\alpha }_{N\mathcal{j}}\right)}\) represents the internal saturation power level of the amplifier, hν is the photon energy at \( {{\uplambda }}_{\mathcal{j}},\)\( \nu =\frac{c}{{{\uplambda }}_{\mathcal{j}}}\), \( A\) is the cross section of the active region, \( {\tau }_{sc} \)is the spontaneous lifespan of a carrier, ϵ is the mode confinement factor and differential gain is denoted as \( {\alpha }_{N\mathcal{j}}\). Unsaturated single pass gain denoted by \( \stackrel{-}{{g}_{\mathcal{j}}^{m}} \)at \( {{\uplambda }}_{\mathcal{j}}\) is expressed as Eqs. (8),

$$ \stackrel{-}{{g}_{\mathcal{j}}^{m}}=exp\left(?{\alpha }_{\mathcal{j}}\left(\stackrel{-}{N}\right){L}^{m}\right)$$
(8)

where, N denotes the unsaturated carrier density and mth stage SOA length is represented by \( {L}^{m}\). Considering the key significance of nonuniform gain saturation and spectral susceptibility of unsaturated gain in XGM, the parameter \( \gamma =\frac{{\alpha }_{N1}}{{\alpha }_{N2}}\) explores the variance in differential gains at \( {\lambda }_{s}\) and \( {\lambda }_{c}\). Normalized input probe power \( {\mathcal{x}}_{20}^{1}\), \( \stackrel{-}{{g}_{1}}\), \( \stackrel{-}{{g}_{2}}\), and \( \gamma \), are parameters that need to be solved which can be measured on employed SOA easily.

Exploiting above functions, the output extinction ratios (ER) for both modulated pump and CW probe signals at each stage are expressed as in Eqs. (9) and (10),

$$ E{R}_{o,1}^{m}={x}_{10}^{m}\left(1\right){g}_{1}^{m}\left(1\right)/{x}_{10}^{m}\left(0\right){g}_{1}^{m}\left(0\right)$$
(9)
$$ E{R}_{o,2}^{m}={x}_{20}^{m}\left(1\right){g}_{j}^{m}\left(1\right)/{x}_{20}^{m}\left(0\right){g}_{2}^{m}\left(0\right)$$
(10)

where, \( {g}_{1}^{m}\left(0\right)={g}_{1}\left({x}_{10}^{m}\left(0\right)\right)\), \( {g}_{1}^{m}\left(1\right)={g}_{1}\left({x}_{10}^{m}\left(1\right)\right)\), \( {g}_{2}^{m}\left(0\right)={g}_{2}\left({x}_{10}^{m}\left(1\right)\right)\) and \( {g}_{2}^{m}\left(1\right)={g}_{2}\left({x}_{10}^{m}\left(0\right)\right)\) can be solved using Eqs. (5) and (6), respectively.

Furthermore, signal to noise ratios (SNR) of the idler in relation to the incoming signal at each stage can be expressed as in Eqs. (11),

$$ SN{R}^{m}=\frac{{\left(S\left(1\right)-S\left(0\right)\right)}^{2}}{{\left(\sqrt{{N}_{t}^{m}\left(0\right)}+\sqrt{{N}_{t}^{m}\left(1\right)}\right)}^{2}}$$
(11)

where, \( S\left(0\right)\) and \( S\left(1\right)\) are the received signal current of the probe for the ‘0’ and ‘1’ bits respectively, then exploiting this we can express the bite error rate (BER) as the Eqs. (12),

$$ BE{R}^{m}=\frac{1}{\sqrt{2\pi }}\left(\frac{\text{e}\text{x}\text{p}(-\frac{SN{R}^{m}}{2})}{\sqrt{SN{R}^{m}}}\right)$$
(12)
Fig. 3
figure 3

Experimental setup. PC: Polarization controller. PBC: Polarization beam combiner. PBS: Polarization beam splitter, OBPF: Optical band pass filter, S/P: Serial-to-parallel converter, LO: Local Oscillator, PC: Polarization controller, VOA: Variable optical attenuator

Full size image

3 Device Description and Operation

The experimental configuration of Indium Phosphide photonic integrated all optical wavelength conver of 128 GBaud DP-16QAM signals using XGM in SOAs is shown in Fig. 3. Input data signal is labeled as CW 1 and continuous (CW) probe beam are labelled as CW 2. A 128 GBaud DP-16QAM signal is created using a CW laser operating at central wavelength 1550 nm. This laser drives a DP-16QAM modulator, which is in turn stimulated by a \( {2}^{31}-1\) pseudo random bit sequence (PRBS) generator. The electrically driven optical modulator is designed employing serial-to-parallel converters, electrical biasing, Mach-Zehnder modulators (MZMs), phase shifters, electrical gain, couplers, pulse shapers, PRBS generator, PBC and PBS. Electrical gain is 2 and electrical biasing is 1.5 for the transmitter. This input signal (pump) and probe signal’s polarization states are adjusted by the polarization controller (PC) to bring them into alignment with the transverse electric mode of the SOA waveguide. Polarization multiplexed emulator for dual polarization is used after a 128 GBaud DP-16QAM transmitter to modify timing and for decorrelation, include a time delay. After compensating the attenuation, variable optical attenuators (VOAs) are incorporated to adjust the power variation of the two separated branches caused by arbitrary polarization. This configuration consists of two erbium doped fiber amplifiers (EDFAs), filters, couplers, InP-PIC, DP-16QAM receivers, polarization beam splitters (PBSs), polarization beam combiners (PBCs), digital signal processing (DSP) unit, decision, and QAM demodulator. To preserve the OSNR, EDFA’s noise figure is set at 4 dB. To withstand the amplified spontaneous emission (ASE) noise generated by the EDFA, an adjustable band pass filter is included. The input data (CW 1) and probe (CW 2) have frequencies of 1550 nm (\( {\lambda }_{s}\)) and 1540 nm (\( {\lambda }_{c})\), respectively. After that, a 3dB coupler is used to combine the strong pump signal with probe signal CW 2 at wavelength \( {\lambda }_{c}\). The photonic integrated circuit (PIC)’s input optical power is set to 0 dBm using EDFA, BPF, VOA, and PC. The optical band pass filter, which is subsequent by the PIC wavelength converter, receives the output from the 3 dB coupler. Two cascaded nonpolarized SOAs coupled with a delay line interferometer (DLI) filter and a adjustable band pass filter (BPF) compose a PIC wavelength converter. 150 mA of current is delivered to SOA 1 and SOA 2. In addition, there is a 4.5 dB insertion loss because tapered fibers are employed at the beginning and end of the PIC for light injection and collection to and from the chip. There are two stages to cross-gain modulation when the probe signal is coupled into the PIC. By taking use of the pump signal at stage 1, the probe signal undergoes XGM in SOA 1. As illustrated in Fig. 2., it then undergoes self-gain modulation in SOA 2 as a result of the pump signal’s rising attenuation after going through the DLI filter. To reduce the ASE noise, another optical band pass filter (OBPF) is used. A DP-16QAM receiver receives the converted signal (idler), after which a decision-making unit integrated with DSP module, and bit error rate test set are used to assess the performance of the proposed system. All of the necessary parameters for the suggested wavelength conversion are displayed in Table 1 for the computation.

Table 1 Computational variables for the wavelength conversion
Full size table

4 Results and Discussion

The experimental arrangement depicted in Fig. 3 and the computational parameters outlined in Table 1 guide our investigation. First, we have traced the output spectrum of the 128 GBaud DP-16QAM signals shown in Fig. 4 at a central frequency of 1550 nm. DP-16QAM modulator is fed with 0 dBm input power and received power at its central frequency is -15 dBm. To ensure optimal performance of the all-optical wavelength conversion (AOWC) system, we maintain the probe signals within their 3 dB bandwidth range, aligning the pump wavelength with the minima of the Dispersion-Less Interferometer (DLI) filter. This arrangement effectively reduces the pump power via the DLI filter just before it reaches SOA 2. To expedite gain recovery and reconfiguration, self-gain modulation within SOA 2 is implemented. An optical spectrum analysis is conducted at the output of the Photonic Integrated Circuit (PIC) immediately prior to filtering by the Optical Bandpass Filter (OBPF), as shown in Fig. 5, revealing the presence of the two output wavelengths and the associated Four-Wave Mixing (FWM) components. Shift in wavelength is observed either side of the central frequency. These two converted wavelengths are down-converted idler and up-converted idler at 1540 nm and 1555 nm, respectively. The output spectrum is comprised of sampled, parameterized, and noise components. The noise power level is represented by the green color, while the signal power is represented by the blue cooler. By raising the OSNR of transformed wavelength signals, noise can be decreased. As a result, more pump power and lower probe power are needed to increase OSNR. It has been noted that AOWC performance is limited by increasing probe power. With 40 CPE symbols per block, the DP-16QAM receiver’s dark current and responsivity are kept at 10 nA and 1 A/w, respectively. To manage the laser operation point, circuit should include more VOA at the laser output together with a tunable laser in order to feasibility of the conversion in entire C-band.

Figure 6 shows the bit error rate (-log (BER)) curves for varying power of the 128 GBaud DP-16QAM signal. Received power at the output of PIC (measured on a test SOA) includes all coupling and chip losses. Integrated SOAs on PIC are driven by 150 mA injection current. It is essential to maintain the SOA at a lower biasing to address laser instability issues. The BER is evaluated across different received power levels for the input back-to-back (BTB) signal (at 1550 nm), the down-converted signal (at 1540 nm), and the up-converted signal (at 1555 nm).

Fig. 4
figure 4

Optical Spectrum at the output of 128 GBaud DP-16QAM Signal at signal frequency 1550 nm and input power 0 dBm

Full size image
Fig. 5
figure 5

Optical Spectrum of idlers (converted wavelengths) at the output of the PIC.

Full size image

Increased probe input power leads to the escalation of nonlinear effects such as self-phase modulation (SPM), cross-polarization modulation (XPolM), and cross-phase modulation (XPM). This elevation in nonlinearities results in a decrease in four-wave mixing (FWM) power, consequently affecting the performance of idler and back-to-back signal (BTB) bit error rate (BER). When comparing with the BTB signal at 128 GBaud, it was observed that power penalties at a BER of 10− 9 decrease to less than 1 dB and 2 dB for down-converting and up-converting signals, respectively. An anticipated drawback of wavelength conversion based on cross-gain modulation (XGM) is the wavelength tilt effect in a saturated semiconductor optical amplifier (SOA), potentially leading to a notably higher power penalty for up-conversion due to reduced receiver sensitivity. The optical-to-signal noise ratio (OSNR) experiences a drastic reduction due to increased absorption losses in the SOA. Enhanced signal quality is observed for larger OSNR (> 40 dB) as BER decreases.

The difference in received power between the 1540 nm and 1555 nm BTB signals is attributed to the wavelength dependence of the employed meter’s sensitivity. No error-floor is observed for the converted wavelengths (multicasting), as indicated in Fig. 6. BER exhibits a sharp decrease with received power within the 0 dBm range; beyond 0 dBm, the BER curve flattens. In an effective all-optical wavelength conversion (AOWC) system, the pump signal power should exceed the launched power of the probe input signal. The precision of measurements, influenced by the bandwidth limitations of the modulator and photodiode, impacts the linearity of BER curves on a logarithmic scale. For the second semiconductor optical amplifier (SOA 2), the optimum working bias is 150 mA, with a transparency criterion of 55 mA. SOA 2 notably offsets the sluggish recuperation period of SOA 1. As a common consequence of cross-gain modulation (XGM) in SOAs, up-conversions are susceptible to the asymmetric gain tilt effect. Utilizing extended SOAs of 1.5 mm helps mitigate this penalty. The received power in optical communication systems includes both the signal power that carries the information and the various noise sources that exist in the system. SNR has a direct influence on received power. The resultant of signal and noise power is mathematically represented as received power. As a result, BER measures system performance while accounting for the noise caused by additional nonlinearity in the wavelength conversion. Because of the nonlinear media, the BER of up- and down converted signals increased slightly after wavelength conversion when compared to the BTB signal, as shown in Fig. 6. BER decreases vertically downwards as– log (BER) is increasing as a function of received power.

Increased pumping power, retention of the polarization state, and immunity to orthogonality loss can all help to lessen the BER degradation. In order to lower the optical signal to noise ratio and stimulated Brillouin scattering (SBS), polarization multiplexing is utilized. To minimize the power penalties, DSP techniques are utilized to adjust the power-dependent Kerr nonlinearities. Phase conjugation, digital back propagation, nonlinearity equalization, and polarization maintenance are DSP techniques.

Fig. 6
figure 6

The -log (BER) versus received power for input signal, down converted signal and up converted signal

Full size image

The comparison between the signal output power with the pump activated (pump ON) and without the pump activated (pump OFF) defines the ON-OFF gain. In Fig. 7, the gain saturation characteristics are depicted concerning the input signal power of a DP-16QAM signal for two different output wavelengths. The input data signal operates at a wavelength of 1550 nm, with its down-converted and up-converted wavelengths at 1540 nm and 1555 nm, respectively. Notably, when the input signal is at 1550 nm, a saturation in gain occurs within a 1 dB margin between the original and converted signals for both 5 dBm and 10 dBm probe powers. Specifically, at input powers below − 5 dBm, a 15 dB ON-OFF gain is observed for 5 dBm probe power, while a 25 dB ON-OFF gain is recorded for 10 dBm probe power. Signal 2 exhibits significant amplification compared to signal 1, aligning with expectations for enhanced performance at input power levels below − 5 dBm. The range of input signal power for ON-OFF gain measurement spans from − 30 dBm to 15 dBm.

Fig. 7
figure 7

A comparison of the gain saturation of the DP-16QAM signals and wavelength converted signals as a function of input signal power. Signal 2 (at 10 dBm probe power) is amplified signal compare to signal 1 (at 5 dBm probe power)

Full size image
Fig. 8
figure 8

The output extinction ratio (ER) for PIC integrated AOWC as a function of the received input power

Full size image

Figure 8 shows the on-chip input power dependence of the output extinction ratio (ER) for the input signal (BTB) and its converted wavelengths. The definition of ER is the ratio of a digital signal’s greater optical level to its lower optical power level. Input wavelength is 1550 nm and the output of AOWC is 1555 nm and 1540 nm for both up conversion and down conversion respectively. Extinction ratio improves gradually as the input power increases to the receiver, and then level off at 15.2 dB, 14.5 dB and 12.95 dB, respectively, for back-to-back (BTB), down conversion and up conversion respectively. Extinction ratio starts decreasing after reaching at peak, at 5 dBm power. The shape of the extinction ratio curve is consequence of the receiver saturation effects and electro-absorption modulators (EAM). The other factor that significantly affects the ER is receiver pre amplifier response. The optical gain from the receiver is degraded as input power approaches the amplifier saturation power.

Fig. 9
figure 9

The output extinction ratio (ER) for PIC integrated AOWC in relation to the converted wavelength

Full size image

Figure 9 displays the wavelength dependence of the extinction ratio for different level of probe powers. It is observed that ER increases with growing wavelength upto 1550 nm. The extinction ratio is sharply degraded as the wavelengths of a converted signal move to a shorter side, higher extinction ratio is obtained in longer wavelength range. It should be noted that extinction ratio is higher for 10 dBm probe power at shorter wavelength (< 1545 nm) and extinction ratio is higher for 5 dBm probe power at wider wavelength (> 1545 nm). This indicated that input power has to be optimized for the different converted wavelengths. The changes in extinction ratio is affected by the differential gain with the difference of signal and probe wavelebgths.

Figure 10 shows the varaiation of conversion efficiency (CE) for a range of converted wavelength, in presence and absence of pumping. The ratio between the output power of the signal without a pump and the output power of the converted wavelength is referred to as conversion efficiency. CE is measured over the 100 nm wavelength conversion range (1500 nm to 1600 nm). More than 60 nm frequency shift (wavelength conversion range) is obtained for a saturation level of conversion efficiency. At 1550 nm, highest conversion efficiency of converted signal is -13 dB for signal without pumping using SOA and conversion efficiency of other formats is compared with respect to this. At 1550 nm, highest conversion efficiency is -5 dB for signal without pumping using InP-PIC (improved by 8 dB), -2 dB for converted signal with pumping using SOA (improved by 11 dB), and − 1 dB for converted signal with pumping using InP-PIC (improved by 12 dB). So It is concluded that CE is improved by 8 dB, 11 dB and 12 dB for InP-PIC signal without pumping, SOA signal with pumping, InP-PIC signal with pumping, respectively, with respect to SOA signal without pumping. It’s recommended to lower the average signal power to enhance conversion efficiency rather than escalating the probe power, which also narrows the output power range. Additionally, balancing between output extinction ratio and conversion efficiency is necessary. To achieve greater conversion efficiency and minimize power penalties, the launched signal strength should ideally fall within the range of -25 to -5 dBm relative to the probe signal power (5 dBm to 20 dBm).

Fig. 10
figure 10

The correlation between the wavelength conversion range and conversion efficiency, with and without pumping

Full size image

The eye diagram produced at the output of AOWC is displayed in Fig. 11 which reflects clear eye opening without any distortion. Eye diagram is measured at -10 dBm input power, it is evident that slow recovery time at stage 1 (SOA 1) has been compensated drastically at stage 2 (SOA 2). Increasing biasing current of SOA 2 from 60 mA to 150 mA leads to a wide-open clear eye with less error. This indicates that good conversion performances are met with proposed implementation. To control the spurious reflection from the circuit and laser operation point, we required to employ additional VOAs.

Fig. 11
figure 11

Eye diagram at the output of PIC at 150 mA bias current of SOA 2

Full size image

In brief, the XGM implementation is easy to accomplish, polarization independent due to polarization independent boosters, and it has less power penalty at higher data rate. A Correlative analysis of some of the recent research investigations is presented in comparison to the proposed work, as shown in Table 2.

Table 2 Comparison between current research and previous studies on wavelength conversion
Full size table

5 Conclusion

We have demonstrated the theoretical analysis and experimental implementation of an Indium Phosphide (InP) photonic integrated wavelength conversion of 128 GBaud DP-16QAM signals using cross-gain modulation through semiconductor optical amplifiers. To enhance wavelength conversion efficiency, advanced modulation method DP-16QAM is employed to mitigate the SBS. is suppressed using the advanced modulation method DP-16QAM. Throughout the 60 nm wavelength conversion range, conversion efficiency is improved by 12 nm. Degradation in the extinction ratio is efficiently addressed by dual cascaded SOAs. Extinction ratio measurements across a 35 nm wavelength span (from 1525 nm to 1565 nm) yield values of 15.2 dB (back-to-back), 14.5 dB (down-conversion), and 12.95 dB (up-conversion) at a 128 GBaud rate.

For up-converted and down-converted signals, power penalties are lowered to less than 1 dB and 2 dB, respectively, at a BER of \( {10}^{-9}\). BER is seen in the received power range of -12 dBm to 4 dBm. Both up- and down-conversion at greater OSNR have resulted in a decrease in the BER of the converted wavelength. Additionally, at less than − 5 dBm input power, substantial ON-OFF gain is achieved: 15 dB ON-OFF gain is recorded at 5 dBm probe power, and 25 dB ON-OFF gain is realized at 10 dBm probe power. Input signal power and CW probe power has to optimized to reduce the power penalty and enhance the extinction ratio. Biasing and injection current of SOAs play a vital role in performance of AOWC, we have considered 150 mA injection current. An agreement between yield extinction ratio and CE has to be made. Our experimental findings correlate well with theoretical analysis as expected. Thus, this implementation ensures a potential solution for next generation inter-networks of data centers.