Robust Dual-Wavelength DRoF Link by Quasi-2-bit 256-QAM OFDM With Delta-Sigma Modulation

We experimentally demonstrated a robust dual-wavelength digital radio-over-fiber (DRoF) link by quasi-2-bit delta-sigma modulated 256- quadrature amplitude modulation orthogonal frequency-division multiplexing (QAM OFDM) data. By optically constructing the 2-bit delta-sigma modulated data with the optical group delay compensation technique, the proposed system combined the advantages from either the conventional 2-bit or 1-bit system. Due to the benefits of higher signal-to-noise ratio and robust anti-nonlinearity to the optical links, the proposed approach performed the widest dynamic ranges of 25.78, 17.81, and 13.46 dB after 25-km single-mode fiber (SMF) transmission by using the 400-MHz 16-, 64-, and 128-QAM OFDM data, respectively. These values were much better comparing with either the conventional 2-bit or 1-bit system. Up to 256-QAM OFDM data with 400-MHz bandwidth was supported by the proposed system for transmitting over 25-km SMF at the lowest overall receiving power of -10.72 dBm. This work provided a good solution for the beyond 5G or 6G DRoF link applications.


I. INTRODUCTION
I N THE 5th generation (5G) wireless network, centralized radio access network (C-RAN) in combination with radioover-fiber (RoF) link has been regarded as the indispensable approach for providing high-speed Internet connection with low-latency and ubiquitous mobile coverage [1].As the wireless network evolves into 6th generation (6G) or beyond 5th generation (B5G), smaller cells and new spectra will be adopted according to the B5G/6G roadmaps [2], [3], [4].The cell sizes are typically reduced because of the demand for enhanced data rates and the increasing users/user densities, which also leads to the reduction of transmit powers [5], [6], [7], [8].The frequency bands of the utilized radio frequency (RF) signals are upshifted to millimeter-wave (mm-wave) or sub-THz bands for wider available bandwidths [9], [10], [11], [12].In this scenario, transmitting the RF signals from the baseband units (BBUs) to numerous remote radio heads (RRHs) in a cost-effective and robust way becomes a significant issue for the B5G/6G RoF links.For the RoF implementation, various approaches have been demonstrated the diversity and availability for the combination of high-speed network and wireless connection [13], [14], [15], [16].Comparing with the analog RoF (ARoF), digital RoF (DRoF) were regarded as the more suitable solution because the digital-to-analog convertor (DAC) was shifted from BBUs to RRHs, making the n-bit digitized signal transmit over the optical links [17].By such an implementation, the impact of nonlinearity from the used optical devices could be significantly relieved in this approach [18], [19], [20].However, several drawbacks need to be solved for the traditional DRoF links in the B5G/6G scenario.Firstly, the constructing cost of numerous RRHs will raise tremendously because the DAC components in all RRHs should cover the high frequency bands ranging from mm-wave up to sub-THz.Furthermore, complicated digital signal processing (DSP) circuits needed to be employed in every RRHs, which was also an obstacle to cost-effective implementations for traditional DRoF links in B5G/6G [17], [21].In such a circumstance, the delta-sigma modulation based DRoF links were proposed [22], [23], [24], which could be an appropriate solution to reach cheapness and robustness simultaneously for the following reasons.For cheapness, only one simple lowpass or bandpass filter needs to be set in the RRH, which means that the complex and expensive DAC components could be waived [17], [21], [25], [26], [27], [28], [29].For robustness, because the delta-sigma analog-to-digital convertor (ADC) only requires 1 or 2 quantization bits typically, the anti-nonlinearity to the used devices in the optical links can be improved.However, different pros and cons would affect the DRoF link when adopting either 1-bit or 2-bit delta-sigma ADC.Theoretically, the 2-bit data could achieve higher signal-to-noise ratio (SNR), because the quantization noise (QN) generated in the 2-bit delta-sigma ADC had lower power than that generated in the 1-bit one [27], [28].Nevertheless, the 1-bit data equipped with better anti-nonlinearity in the optical links comparing to the 2-bit ones.This meant that the 2-bit delta-sigma modulation based DRoF link did not outperform at all conditions.Hence, to grab the advantages from either 2-bit or 1-bit data, we proposed the quasi-2-bit delta-sigma modulation based DRoF link with dual-wavelength optical carrier [25], [26], [28], [29].Fig. 1 illustrates the scheme diagram of the traditional 1-bit (Fig. 1(a)), 2-bit (Fig. 1(b)) and the proposed quasi-2-bit (Fig. 1(c)) deltasigma modulated waveforms with their pros and cons.From Fig. 1(c), the 2-bit delta-sigma modulated quadrature amplitude modulation orthogonal frequency-division multiplexing (QAM-OFDM) data could be optically constructed with 1-bit scheme, which combined the all the advantages and excluded all the drawbacks from either 1-bit or 2-bit system.
In this work, we experimentally demonstrate a robust quasi-2-bit delta-sigma modulated 256-QAM OFDM based dualwavelength DRoF link.This is the extended work from our previous studies [28], [29].To deal with the timing asynchronization issue, the optical group delay compensation technique will be conducted in the proposed DRoF approach.For comprehensive discussions, the conventional 1-bit and 2-bit system will also be employed.During the experiment, the dynamic ranges of each system will be measured and compared by the demodulated constellation plots, error vector magnitudes (EVMs), average SNRs, and the derived bit error rates (BERs).This work is aiming to provide a robust and cost-effective solution to facilitate the B5G/6G networks.The article structure can be divided as follows.Section I is the introduction and background of the delta-sigma modulation based DRoF links.Section II states the working principle and the experimental setup of the proposed DRoF link and the conventional systems.Section III includes the discussions of the optical group delay compensation to the proposed system (part A) and the comparison between the proposed system and the conventional ones (part B).Besides, the transmission feasibility of the 3.2-Gbit/s quasi-2-bit delta-sigma modulated 256-QAM OFDM through the proposed system is also included in this part.Finally, Section IV concludes this work.

II. WORKING PRINCIPLE AND EXPERIMENTAL SETUP
Fig. 2 depicts the testing system of the robust dual-wavelength DRoF link with delta-sigma modulated QAM-OFDM.Two approaches were employed in the transmitter part: the proposed quad-level DRoF link (approach A) and the conventional 1-bit or 2-bit system (approach B), where different kinds of delta-sigma modulated QAM-OFDM data streams were employed.For the testing signal generation, the original binary data sequence was firstly executed the M-ary QAM mapping with gray coding [30].The complex numbers were then filled into a matrix in frequency domain for inverse fast Fourier transform (IFFT).Before IFFT, the Hermitian symmetry matrix was added in advance to exclude the imaginary part.The cyclic prefix (CP) with a ratio of 1/32 was included after IFFT, and then the parallel-to-serial process turned the data into pure real sequence.Later, the training sequence with a ratio of 1/32 was combined with the generated data sequence, where the pure real M-ary QAM-OFDM data streams were ready for intensity modulation or further operation.During the process of delta-sigma modulation, the generated M-ary QAM-OFDM data stream were modulated by a 4th order 1-bit or 2-bit delta-sigma modulator.For the proposed approach A, only 2-bit delta-sigma modulator was employed.To transmit the generated data stream in the on-off keying (OOK) like 1-bit scheme for better anti-nonlinearity, a 1-bit splitter was added to separate the 2-bit M-ary QAM-OFDM data into two 1-bit data streams.The two 1-bit data streams were recognized as most significant bit (MSB) and least significant bit (LSB), which could be independently encoded to different optical carriers.The FFT size of the data was 1024, and the data bandwidth was set as 400 MHz for meeting the 5G standard from 3GPP Release 15 [31].All the aforementioned processes were executed by DSP in MATLAB programs.
In the proposed approach A, the 1-bit MSB and LSB data streams were separately output by two arbitrary waveform generators (AWGs, Tektronix 70001A) at sampling speed of 24 GS/s, which meant that the oversampling rate (OSR) was 30.
To prevent the unwanted timing asynchronization between each signal output, the AWGs were calibrated and synchronized by a synchronization board (SYNC, Tektronix AWGSYNC01) in advance.Two electrical amplifiers (Amps, SHF S126A) with two variable attenuators (VAs) were respectively connected to the outputs of the two AWGs to provide enough waveform amplitude and modulation depth for data encoding.Two C-band laser diodes (LDs, santec ECL-200) with optical polarization adjustment were employed as the dual-wavelength data encoder, which were externally modulated by the aforementioned MSB and LSB data streams through two Mach-Zehnder modulators (MZMs, Fujitsu FTM7937EZ).For controlling the optical path and adjusting the optical power of each encoded LD, two pairs of variable optical delay line (VOD) and variable optical attenuator (VOA) were added at the outputs of the two MZMs.In addition, the VODs were free space type with insertion losses of 0.90 dB, which could hardly impact the transmission performances.Later, a 50%-50% optical coupler (OC) was employed to combine the two encoded LDs and generate the dual-wavelength optical carrier.Before SMF transmission, the dual-wavelength optical carrier was amplified by an erbium-doped fiber amplifier (EDFA) for maintaining adequate optical power.Besides, an optical bandpass filter (OBPF) was also employed to filter out the unwanted amplified spontaneous emission from EDFA.Two different fiber lengths were included in this work, which were back-to-back (B2B: 0-km), and 25-km SMF transmission.The transmitted optical two-tone was received by a 12.5-GHz photodetector (PD, Sevensix SSPD-12.5-2) with optical receiving power adjustment by another VOA.Finally, an oscilloscope (Tektronix MSO73304DX) was employed to sample the quasi-2-bit delta-sigma modulated QAM-OFDM data at a sampling speed of 100 GS/s.The captured data were offline analyzed by another MATLAB program, where the out-of-band noise was firstly filtered out by a low-pass filter, and the recovered QAM-OFDM data was then demodulated.During the demodulation process, constellation plots with the EVMs, SNRs, and BERs were calculated for evaluating the system performances.
In the conventional approach B, all the aforementioned receiver part and the demodulation program were kept the same, while only the transmitter part was rearranged.The same AWG with the combination of Amp and VA was employed to output the 1-bit or 2-bit delta-sigma modulated QAM-OFDM data.Besides, due to the conventional testing system, the 1-bit splitter was excluded in the data generation process, which meant that the 2-bit data streams kept their original waveforms.An optical polarization-adjusted single-wavelength C-band LD was adopted as the optical carrier, which was externally encoded by the aforementioned signals through an MZM.For fair comparison, a VOD and a VOA were connected to the output of the MZM.Based on the testing system illustrated in Fig. 2, the proposed approach A and the conventional approach B were compared in the following sections.
Fig. 3(a) depicts the time domain waveform of the generated 2-bit OFDM (green) and the two 1-bit data streams (blue MSB and red LSB) after the 1-bit splitter output.In the proposed dual-wavelength DRoF system, the MSB and LSB signal were encoded to the 1552.12-and1552.73-nmsingle-wavelength optical carriers, respectively, which were shown in the optical spectrum in Fig. 3(b).To maintain the amplitude proportion of the MSB and LSB waveforms shown in Fig. 3(a), the peak powers of the dual-wavelength optical carrier were respectively adjusted to -8.9 (MSB carrier) and -12.2 dBm (LSB carrier), which were also shown in Fig. 3(b).By this operation, the 3.3-dB power difference maintained the V pp ratio between MSB and LSB as 100% and 50%.Even though some tolerance ranges of optical power difference could be allowable according to our previous work [25], we fix this value in this manuscript for achieving the best transmission performances.In addition, the wavelength spacing between the optical two-tone was arranged to 0.61 nm to prevent the unwanted four-wave-mixing effect after PD detection.The 2-bit OFDM waveform could be recovered when PD detected the dual-wavelength optical carrier simultaneously.

A. Optical Group Delay Compensation
According to our previous studies [25], [26], [28], even though the optically constructed quasi-2-bit delta-sigma modulated QAM-OFDM presented good performances, the signal qualities were limited by the timing asynchronization due to the use of the dual-wavelength optical carrier.The arriving time of each carrier would be different after long fiber propagation.Hence, relieving the impact of this issue should be the first step to construct the robust dual-wavelength DRoF link.Based on the proposed system shown in Figs. 2 and 3(b), the MSB carrier would propagate faster than the LSB carrier due to the shorter wavelength with experienced lower refractive index.In other words, it was reasonable to add additional optical path to the MSB carrier, which made the arriving time the same for both carriers after long fiber transmission.
Fig. 4 shows the EVM performances and the corresponding constellation plots of the demodulated quasi-2-bit delta-sigma modulated 2.4-Gbit/s 64-QAM OFDM (data bandwidth of 400 MHz) with additional path to MSB carrier.The X-axis in Fig. 4 could be recognized as the additional optical path for MSB carrier (shorter wavelength) comparing with the LSB carrier (longer wavelength).For the B2B transmission case, the transmitted quasi-2-bit delta-sigma modulated 2.4-Gbit/s 64-QAM OFDM exhibited the lowest EVM of 3.73% at the condition of no additional optical path adjustment (0-mm additional optical path for MSB carrier).Either adding or subtracting the additional optical path for MSB carrier degraded the EVM performances.For example, the EVM deteriorated to 5.43% and 7.66% when adjusting the MSB carrier additional optical path to −30 or 40 mm, respectively.This meant that the optical two-tone arrived simultaneously without fiber transmission, and even slight optical path difference made asynchronization to the dual-wavelength optical carrier.
After 25-km SMF propagation, the EVM drastically deteriorated to 10.38% for the transmitted quasi-2-bit delta-sigma modulated 2.4-Gbit/s 64-QAM OFDM data without optical path adjustment, which meant that the timing asynchronization between each optical carrier became a strong issue to limit the signal performance.To overcome this barrier, increasing the additional optical path for the MSB carrier optimized the quality of the transmitted signal.For example, the EVM value was optimized to 9.18% when adding 80-mm additional optical path to MSB carrier, and the magnitude reached the best performance of 5.63% when setting the additional optical path to 110 mm.This meant that the optical group delay was perfectly compensated at this condition after 25-km SMF transmission.In addition, the EVM value inversely deteriorated to 6.07% when adding 120-mm additional optical path to MSB carrier because the LSB carrier arrived earlier than the MSB carrier at this condition.Hence, if the fiber length being shortened, less additional optical path will be required due to the less optical group delay between the two carriers, vice versa.In other words, for a specific length of SMF, system optimization needs to be conducted when applying the proposed system.Fig. 5(a) depicts the demodulated subcarrier SNRs of the corresponding data shown in Fig. 4. Except the pink curve, all the curves denoted a specific optical path adjustment value to the MSB optical carrier in the 25-km SMF transmission cases.The pink curve denoted the subcarrier SNRs for the B2B case, which could be the reference to all the other transmission cases with different values of optical path adjustment.By deducting the subcarrier SNR values of each transmission cases from the B2B reference case, the SNR degradation could be revealed, which are illustrated in Fig. 5(b).
Based on Fig. 5(a) and (b), the SNR degraded more at higher frequency range than that at lower frequency range after 25-km SMF transmission without optical group delay compensation (0-mm case).In other words, the former would be impacted by a severer timing asynchronization than the latter did.By gradually adding the optical path to the MSB carrier (from 0 mm to 80, 100, and 110 mm), the SNR values could be improved because the impact from timing asynchronization was relieved.On the other hand, the SNR values degraded when further adding the additional optical path to the MSB carrier from 110 to 120 mm.This could be attributed to the over-compensated optical group delay, which affected the SNR values at lower frequency range.By appropriately adjusting the additional optical path to the MSB carrier (in this work, 110 mm for the 25-km SMF transmission case), the SNR degradations could be improved to less than 4 dB compared to the B2B case.Consequently, we chose 110-mm additional optical path for MSB carrier as the optimized condition for 25-km SMF transmission case which was also employed in the following investigations.

B. Comparison Between the Robust Quad-Level DRoF Link and the Conventional 1-Bit/2-Bit Systems
To evaluate the superiority and the performance of the proposed robust quasi-2-bit DRoF link, the conventional 1-bit and 2-bit systems depicted in Fig. 2 (approach B) were also investigated by the same testing data.In the following investigation, the operating dynamic ranges for each approach were measured by adjusting the intermediate frequency (IF) input powers to the MZMs, which was realized by tuning the VAs at each Amp outputs.The bandwidth of the testing data was fixed to 400 MHz, and the overall optical receiving powers were fixed to −9 dBm for fair comparison.Besides, the delta-sigma modulated QAM-OFDM data with three different QAM levels (16-, 64-, and 128-QAM) were included in the investigation, which denoted the raw data rates of 1.6, 2.4, and 2.8 Gbit/s.Fig. 6 illustrates the demodulated BERs of the transmitted delta-sigma modulated QAM-OFDM data before (B2B cases) and after 25-km SMF transmission with various IF input powers.Fig. 6(a), (b), and (c) denoted the BER values for delta-sigma modulated 16-, 64-, and 128-QAM OFDM data at B2B case, respectively.After 25-km SMF transmission, the corresponding values were depicted in Fig. 6(d), (e), and (f), respectively.In each subfigure in Fig. 6, the blue curve denoted the BER performance from the proposed robust DRoF link (this work, approach A), where the gray and pink curves respectively denoted the BERs from the conventional 1-bit and 2-bit systems (approach B).By tuning the electrical VAs, the IF input power ranged from -19 to 14 dBm for this investigation.
According to each subfigure in Fig. 6, the demodulated BERs from the proposed system exhibited the best values among all the three approaches.Due to the similar trends in each subfigure in Fig. 6, it was reasonable to take one of them for example.Based on Fig. 6(e), the 2.4-Gbit/s delta-sigma modulated 64-QAM OFDM was applied as the testing data.When choosing the optimized condition (IF input power of 5.00 dBm), BER of 1.92 × 10 −10 could be obtained after 25-km SMF transmission in the proposed system.However, the magnitudes for the conventional 2-bit or 1-bit systems were only 5.56 × 10 −8 and 3.67 × 10 −7 , respectively.It was because that better SNRs could be observed in the proposed system than the conventional ones as the previous prediction shown.When down-tuning the IF input power from 5.00 to −4.00 dBm, the obtained BERs deteriorated to 1.50 × 10 −3 , 7.30 × 10 −3 , and 5.80 × 10 −3 for the proposed system, conventional 2-bit, and 1-bit system, respectively.Among this, only the value obtained from the proposed system met the forward error correction (FEC) criterion (BER of 3.80 × 10 −3 ).This could be attributed to the inadequate data amplitudes in each system.Nevertheless, the proposed system outperformed in the low IF input power region because better SNRs could still be obtained than that in the conventional ones.
On the other hand, when enlarging the IF input power from 5.00 to 12.00 dBm, the BERs deteriorated sharply, failing to meet the FEC criterion.This could be attributed to the nonlinearity effect from the employed MZMs under the over-amplified signal input.In addition, due to the good linearity in the employed MZMs (V π of 3.5 V), the performance difference between the proposed system and the conventional ones could hardly be distinguished.Nevertheless, if adopting the directly modulated laser as the transmitter or replacing the current MZMs with the lower nonlinearity ones, the proposed system would outperform in the high IF input power (high nonlinear distortion) regions.In the case shown in Fig. 6(e), 12.00-dBm input should not be applied to any of the systems.Consequently, the IF input power should be maintained within a specific range, namely dynamic range, for all the three approaches.
For analyzing the dynamic ranges, measuring the input IF power limits for each system to meet the FEC criterion was an indispensable step.Based on all the subfigures in Fig. 6, the "lower limit" for the input IF power (dBm) could be defined as the intersection of each BER curve and the FEC criterion in the lower IF input power range.For the same reason, the "upper limit" for the input IF power (dBm) was the intersection in the higher IF input power range.For example, when applying the 2.8-Gbit/s delta-sigma modulated 128-QAM OFDM as the testing data, the IF input power boundary for the proposed system were −6.89 and 11.03 dBm before fiber transmission.The values changed to −2.95 and 10.51 dBm after 25-km SMF transmission.In this circumstance, the dynamic range of a specific system could be defined as the power difference between the lower and upper limits, which could be obtained by subtracting the lower limit value from the upper limit one.For the same example, the dynamic ranges of the proposed system were 17.92 and 13.46 dB when applying the 2.8-Gbit/s delta-sigma modulated 128-QAM OFDM as the testing data before and after 25-km SMF transmission, respectively.
Table I summarizes the dynamic ranges from the used different approach systems by three QAM levels data (16-, 64-, and 128-QAM OFDM) with and without 25-km SMF transmission.The lower and upper limits for each case are also included in Table I.According to Table I, the dynamic ranges for the proposed system were 32.31 dB by using delta-sigma modulated 16-QAM OFDM data.This value would maintain to 25.78 dB after 25-km SMF transmission.However, the dynamic ranges by using the conventional 2-bit system were only 28.42 and 23.80 dB before and after 25-km SMF transmission with the same delta-sigma modulated 16-QAM OFDM data.This could be attributed to the fragile anti-nonlinearity for the 2-bit conventional system when comparing the proposed robust quasi-2-bit one.Moreover, when applying the 1-bit conventional system with the same original 16-QAM OFDM data, the dynamic ranges would further deteriorate to 27.61 and 23.17 dB before and after 25-km SMF transmission.Even though the conventional 1-bit system could mitigate the impact of nonlinearity from the optical link, the inborn characteristics made it hard to have the same SNRs as the 2-bit system did.This could also explain why the conventional 1-bit system had the worst dynamic ranges among all the three approaches.
Based on Table I, the proposed robust quasi-2-bit system exhibited the highest dynamic ranges within all the three approaches because it combined the advantages and excluded the disadvantages from either conventional 2-bit or 1-bit system.Hence, the proposed system was suitable for transmitting the delta-sigma modulated data with higher QAM level.Fig. 7 illustrated the transmission results of the quasi-2-bit delta-sigma modulated 256-QAM OFDM data by the proposed system.The data bandwidth was kept as 400 MHz for meeting the 5G standard from 3GPP Release 15 [31].Fig. 7(a) shows the subcarrier SNRs with the corresponding constellation plots of the transmitted 3.2-Gbit/s delta-sigma modulated 256-QAM OFDM data before and after the 25-km SMF propagation.The EVM of the received data was 3.62% in B2B case, which degraded to 4.21% after 25-km SMF transmission due to the slight pulse broadening effect in the optical link.Both values met the FEC criterion (EVM of 4.51%).From Fig. 7(a), the SNR level of the 400-MHz data in B2B case was slightly higher than that after 25-km SMF transmission.This could be attributed to the slight pulse broadening effect in the optical link.Nevertheless, the subcarrier SNR curve after 25-km SMF transmission remained flat due to the help of optical group delay compensation mentioned in the previous section.Last but not the least, the transmission feasibility of the proposed system by using the aforementioned data was discussed.Fig. 7(b) demonstrates the demodulated BERs of the transmitted 400-MHz quasi-2-bit delta-sigma modulated 256-QAM OFDM data with different overall receiving optical powers.To meet the FEC criterion (BER of 3.8 × 10 −3 ), the lowest available overall receiving optical powers were −12.10 and −10.72 dBm for the 400-MHz data at B2B and 25-km SMF transmission cases, respectively.A power penalty of 1.38 dB could be observed.
Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.Finally, by using the proposed system, a robust dualwavelength DRoF could be realized with quasi-2-bit deltasigma modulated 256-QAM OFDM.though the singlewavelength conventional systems seem relatively simple, the proposed system is still a competitive approach.In the practical system, to reduce the constructing cost, directly modulated lasers may be employed as the data encoder.In such cases, the cost of the additional high-speed modulator in the proposed approach can be waived.This means that the only additional cost for the proposed system is the VODs (comparing with the conventional ones), which is much reasonable compared to the MZMs.Furthermore, optical sources with weaker linearity such as vertical cavity surface emitting lasers be employed, which means that the advantage of using the proposed system will be more obvious than the conventional ones to deal with the linearity of the used optical devices.In other words, even though employing the proposed system slightly increases the constructing cost compared to the conventional systems, the system performances can be improved to a greater extent for the possible applications.

IV. CONCLUSION
We proposed and experimentally demonstrated a highperformance robust dual-wavelength DRoF link by quasi-2-bit 256-QAM OFDM with delta-sigma modulation.The proposed approach combined all the pros and excluded all the cons of traditional 1-bit or 2-bit delta-sigma modulation.By optical delay compensation with additional 110-mm optical path for MSB carrier, the data quality could be improved to nearly the same as B2B case even after 25-km SMF propagation for the 2.4-Gbit/s quasi-2-bit delta-sigma modulated 64-QAM OFDM.The proposed system exhibited the widest dynamic ranges of 25.78, 17.81, and 13.46 dB after 25-km SMF transmission by using 16-, 64-, and 128-QAM 400-MHz OFDM testing data among all the three approaches, respectively, because it benefitted from the higher SNR and relieved the impact from the nonlinearity in the optical link simultaneously.Finally, we successfully transmitted the quasi-2-bit delta-sigma modulated 256-QAM OFDM signal over 25-km SMF with raw data rate of 3.2 Gbit/s, which met the FEC criterion.Power penalty of 1.38 dB could be observed for the 400-MHz data.The proposed system could be a good solution for the B5G/6G DRoF link.

Fig. 2 .
Fig.2.The experimental setup of the robust dual-wavelength DRoF link with delta-sigma modulated QAM-OFDM and the conventional 1-bit/2-bit system.

Fig. 3 .
Fig. 3. (a) The time domain waveform of the generated 2-bit OFDM (green) and the two 1-bit data streams (blue MSB and red LSB) after the 1-bit splitter output.(b) The optical spectrum of the dual-wavelength optical carrier.

Fig. 4 .
Fig. 4. The EVM performances and the corresponding constellation plots of the quasi-2-bit delta-sigma modulated 2.4-Gbit/s 64-QAM OFDM with different additional optical path to MSB carrier with and without 25-km SMF propagation.

Fig. 7 .
Fig. 7.The (a) subcarrier SNRs with the corresponding constellation plots, and the (b) demodulated BERs of the 400-MHz quasi-2-bit delta-sigma modulated 256-QAM OFDM data by the proposed system with and without 25-km SMF transmission.

TABLE I DYNAMIC
RANGES FOR EACH SYSTEM TO MEET FEC CRITERION