Source-linked AI summary
Single-laser 32.5 Tbit/s Nyquist WDM transmission
David Hillerkuss, Rene Schmogrow, Matthias Meyer, Stefan Wolf, Meinert Jordan, Philipp Kleinow, Nicole Lindenmann, Philipp C. Schindler, Argishti Melikyan, Xin Yang, Shalva Ben-Ezra, Bernd Nebendahl, Michael Dreschmann, Joachim Meyer, Francesca Parmigiani, Periklis Petropoulos, Bojan Resan, Aandreas Oehler, Kurt Weingarten, Lars Altenhain, Tobias Ellermeyer, Matthias Moeller, Michael Huebner, Juergen Becker, Christian Koos, Wolfgang Freude, Juerg Leuthold
TL;DR
High-rate single-laser transmission is of interest because cost and power concerns favor reduced component count. This paper uses real-time sinc-shaped Nyquist WDM and reports 32.5 Tbit/s over 227 km at 6.4 bit/s/Hz.
Problem
High-rate single-laser systems are of interest because reducing component count matters for cost and power consumption.
Source: §I · p. 1Method
The system generates 325 carriers from one laser and uses real-time 64-tap FIR processing to encode sinc-shaped Nyquist pulses without guard bands.
Source: §I · p. 2Results
32.5 Tbit/s was transmitted over 227 km with 6.4 bit/s/Hz net spectral efficiency using a single laser.
Source: §I · p. 2, §VII · p. 8Takeaways & Limitations
The experiment demonstrates 16QAM Nyquist WDM with real-time sinc-pulse shaping and carrier spacing equal to the symbol rate.
Source: §VII · p. 8Back-to-back performance was mainly limited by linear crosstalk associated with the limited number of FIR taps.
Source: §VI · p. 6
Abstract
from arXiv · showhide
We demonstrate 32.5 Tbit/s 16QAM Nyquist WDM transmission over a total length of 227 km of SMF-28 without optical dispersion compensation. A number of 325 optical carriers are derived from a single laser and encoded with dual-polarization 16QAM data using sinc-shaped Nyquist pulses. As we use no guard bands, the carriers have a spacing of 12.5 GHz equal to the Nyquist bandwidth of the data. We achieve a high net spectral efficiency of 6.4 bit/s/Hz using a software-defined transmitter which generates the electrical modulator drive signals in real-time.
I. INTRODUCTION
The paper addresses whether real-time Nyquist pulse shaping can support 16QAM Nyquist WDM over significant distances, using a single laser to reduce component count. It reports 32.5 Tbit/s transmission over 227 km at 6.4 bit/s/Hz.
I. INTRODUCTION
- Single-laser Tbit/s systems are of interest because reducing component count matters for cost and power consumption.Earlier OTDM systems reached 10.2 Tbit/s at 2.6 bit/s/Hz and demonstrated 10.2 Tbit/s over 29 km of dispersion-managed fiber.Source: §I · p. 1
- Prior OFDM work reached 26 Tbit/s over 50 km with dispersion-compensating modules and 5.0 bit/s/Hz net spectral efficiency.The introduction also cites a 600 km, 1 Tbit/s-plus OFDM demonstration at 3.3 bit/s/Hz and a 10.8 Tbit/s super-channel.Source: §I · p. 1
- Sinc-shaped Nyquist pulses have rectangular spectra confined to the Nyquist bandwidth, allowing tight channel spacing; neighboring carriers need no phase control.Earlier Nyquist WDM demonstrations used small guard bands, while a 400 Gbit/s demonstration at 3.7 bit/s/Hz computed its signal offline with a 601-tap FIR filter.Source: §I · pp. 1–2
- Real-time 16QAM Nyquist WDM had not been demonstrated because finite-tap processing must represent acausal sinc pulses, leaving its long-distance feasibility uncertain.The paper identifies this as the outstanding question addressed by its transmission experiment.Source: §I · p. 2
- 32.5 Tbit/s was transmitted over 227 km using 325 carriers from one laser, real-time 64-tap FIR shaping, no guard bands, and 6.4 bit/s/Hz net spectral efficiency.The paper describes this as a record aggregate line rate for a single laser.Source: §I · p. 2
II. BENEFITS AND CHALLENGES OF NYQUIST WDM TRANSMISSION SYSTEMS
Nyquist WDM packs adjacent rectangular channel spectra at symbol-rate spacing, but producing and receiving that signal requires precise spacing and specialized signal processing. The reduced bandwidth also trades against more digital processing and clock-recovery challenges.
II. BENEFITS AND CHALLENGES OF NYQUIST WDM TRANSMISSION SYSTEMS
- True Nyquist WDM uses immediately adjacent, nonoverlapping rectangular spectra whose bandwidth equals the encoded symbol rate; sinc-shaped pulses produce that spectrum.The minimized bandwidth makes efficient use of the electrical bandwidth of DACs, driver amplifiers, and modulators, but increases digital signal processing and requires oversampling for anti-aliasing filters.Source: §II · p. 2
- For neighboring channels with symbol rates R_N and R_N+1, the required spacing is f_Δ = (R_N + R_N+1)/2; equal-rate channels therefore have spacing R.Larger spacing wastes optical bandwidth, while smaller spacing significantly increases linear crosstalk.Source: §II · p. 2
- Receivers coarsely select channels optically and perform final sharp-edged channel selection with digital signal processing.Residual neighboring-channel components after optical filtering are removed by digital brick-wall filtering in the receivers.Source: §II · p. 2
- Nyquist signals lack noticeable spectral components at the sampling-clock frequency, so most standard clock-recovery algorithms do not work.The paper points to its previously presented clock-recovery solution.Source: §II · p. 2
III. NYQUIST WDM SYSTEM CONCEPT
The system concept separates optical-carrier generation from modulation, then combines and splits odd and even channels through filtering and coupling. The architecture uses an optical comb for evenly spaced carriers and digital receiver filtering for residual neighboring-channel spectra.
III. NYQUIST WDM SYSTEM CONCEPT
- An optical comb source provides inherently equidistant carriers, limiting frequency drift that would increase neighboring-channel crosstalk.The proposed source uses a single mode-locked laser, two EDFAs, an HNLF, and a waveshaper; the authors question whether 325 stabilized lasers could match its energy efficiency.Source: §III · pp. 2–3
- The transmitter separates carriers with an interleaver and two wavelength demultiplexers, then encodes data and combines odd and even channels with separate multiplexers and a coupler.The schematic illustrates the transmitter and receiver arrangement for six carriers.Source: §III · p. 3, Figure 1 · p. 3
- The receiver uses a coupler and two optical demultiplexers to separate odd and even carriers before Nyquist receivers.Optical demultiplexers reduce insertion loss and avoid loading balanced detectors with power from unwanted carriers.Source: §III · p. 3, Figure 1 · p. 3
- Residual neighboring-channel spectra remain after optical filtering and must be removed by digital brick-wall filtering in the Nyquist receivers.The optical multiplexers are selected to combine odd or even channels without limiting their Nyquist spectra.Source: §III · p. 3
IV. IMPLEMENTATION OF A NYQUIST PULSE TRANSMITTER
The real-time transmitter uses FPGA-based FIR pulse shaping to generate sinc-shaped Nyquist pulses, then converts and filters the signals before optical modulation. Its 64-tap filter balances pulse shaping against the FPGA’s available logic and processing delay.
IV. IMPLEMENTATION OF A NYQUIST PULSE TRANSMITTER
- Two real-time transmitters generate 16QAM sinc-shaped pulses with two samples per symbol using a 64-tap FIR filter programmed on FPGAs.The PRBS length is 2^15−1, and the number of taps affects processing delay while being limited by available FPGA logic.Source: §IV · p. 3, Figure 2 · p. 4
- The DACs operate at 25 GSa/s to produce 12.5 GBd signals with 6.25 GHz electrical bandwidth.Two DAC types were used because of laboratory availability, with no significant performance difference observed.Source: §IV · p. 3
- After digital pulse shaping, the I and Q signals are converted by DACs, low-pass filtered to remove image spectra, amplified, and fed to an optical IQ modulator.The transmitter figure also shows simulated I/Q eye diagrams, a sinc pulse, and its output spectrum.Source: §IV · pp. 3–4, Figure 2 · p. 4
V. NYQUIST WDM EXPERIMENT
The experiment combines 325 carriers from a mode-locked-laser comb with two Nyquist transmitters and emulated polarization multiplexing. The signal is sent through SMF-28 with EDFA-only amplification and assessed using coherent reception and EVM-derived BER estimates.
Optical comb source
- 325 optical carriers are generated from a mode-locked laser, equalized with a waveshaper, and spaced 12.5 GHz apart.The comb is broadened in highly nonlinear fiber; unstable central lines are replaced with the original laser comb.Source: §V · p. 4
Modulation
- Odd and even carriers are separately modulated at 12.5 GBd, matching the 12.5 GHz carrier spacing, then combined and polarization-multiplexed.A 5.3 ns delay decorrelates the two transmitter outputs before polarization multiplexing is emulated.Source: §V · p. 4
Transmission and reception
- The resulting signal is transmitted over 75.78 km or 227.34 km of SMF-28 using EDFA-only amplification, with optimum total launch power of 18 dBm.After transmission, a waveshaper selects the carrier of interest for coherent reception and signal processing.Source: §V · p. 4
System overview
- The setup uses a coherent Nyquist WDM receiver and emulates polarization multiplexing; polarization mode dispersion is not compensated.The figure depicts the comb generation, odd/even modulation, SMF-28 transmission, and coherent receiver.Source: §V · pp. 4–5, Figure 3 · p. 5
Performance measurement
- Signal quality is characterized by EVM, with BER estimated from EVM data and checked by direct BER measurements at selected points.The BER measurements cover back-to-back and 227.34 km transmission for carriers selected for higher EVM, providing an upper limit.Source: §V · p. 5
VI. EXPERIMENTAL RESULTS
Across back-to-back, 75.78 km, and 227.34 km measurements, EVM remained below the stated limit for a BER of 1.8×10−2. The experiments also examine neighboring-channel crosstalk and check whether EVM-derived BER estimates agree with measured BER.
VI. EXPERIMENTAL RESULTS
- At 227.34 km, total EVM rose 1.7 percentage points from the 10.3% back-to-back reference, while all carriers remained below the limiting EVM for BER = 1.8×10−2.At 75.78 km, total EVM rose 1.0 percentage point.Source: §VI · p. 5, Figure 4 · p. 6
- Non-ideally gain-flattened EDFAs contributed to uneven received spectra; the authors expected better gain flattening to significantly increase achievable transmission distance.Source: §VI · p. 6
- At 1564.58 nm, BER increased from 1.4 × 10−4 without neighboring channels to 2.1 × 10−3 with them; the authors attribute back-to-back performance limits mainly to linear crosstalk.At 1563.66 nm, BER rose from 6.3 × 10−4 without neighboring channels to 1.2 × 10−3 with them; residual crosstalk degraded EVM by 1.5 to 2 percentage points.Source: §VI · p. 6
- Calculated BER differed only slightly from measured BER, supporting the BER–EVM relationship based on a Gaussian-noise hypothesis.The I- and Q-error-vector analyses support a Gaussian distribution of added noise.Source: §VI · p. 7, Figure 5 · p. 7
- A 64-tap FIR filter suffices to implement Nyquist WDM transmission, according to the summary of results.Source: §VI · p. 7
- The authors caution that reception quality could not be compared with 16QAM WDM experiments because neighboring-carrier linear crosstalk limited EVM, or with experiments using many separate lasers.Source: §VI · p. 8
VII. CONCLUSION
The paper presents 16QAM Nyquist WDM as a promising next-generation approach, reporting 32.5 Tbit/s aggregate data rate and 6.4 bit/s/Hz net spectral efficiency from a single laser.
VII. CONCLUSION
- 32.5 Tbit/s aggregate data rate and 6.4 bit/s/Hz net spectral efficiency were achieved, setting a new data-rate record for a single laser source.Source: §VII · p. 8
- The paper demonstrates 16QAM Nyquist WDM for the first time with symbol rate equal to carrier spacing and real-time digital sinc-pulse shaping.Source: §VII · p. 8
- Nyquist WDM improves spectral efficiency and transmission distance compared with previously investigated all-optical OFDM systems.Source: §VII · p. 8