Innumerable great achievements were accomplished in the development process since 1976. For decades since the first practical optical fiber of China was developed at Wuhan Research Institute of Posts and Telecommunications(predecessor company of FiberHome), FiberHome has been striving forward, breaking bottlenecks in the optical communication field one after another and leading the industry in technological development.

Looking into the future at a new starting point in 2023, FiberHome points out ten challenges facing the development of optical communication and looks forward to scaling new heights in optical communication through joint innovation with operators, customers, and partners in the industry.

Challenge I Can the “Moore's Law” of optical communication continue? How will it develop in the future?

Since WDM systems were put into commercial use, each time the speed upgrading of backbone optical transmission followed the law of “the transmission distance is proper and the system capacity linearly doubles with the speed upgrading”, which has been the conventionalized “Moore's Law” of optical communication in this industry. For the long-distance transmission of 400G WDM systems, the application of high-bandwidth devices and C- and L-bands has solved the difficult problem of increasing system capacity for long-distance transmission. In the future, how system capacity can be further increased for long-distance transmission will be a focus of research in the industry. To speed up the single-wave velocity to 800G and super high rate above 1T, 200 GBaud+ higher bandwidth devices are needed and the channel spacing needs to be widened to 200 GHz or more. The 80-wave multiplexing requires broader spectrum resources beyond C+L band. It is a major challenge to break the dilemma of “increasing speed without increasing capacity”. To increase the number of wavelengths in multiplexing through broadening spectrum when the single-wave velocity remains at 400 Gbit/s, a series of problems related to laser light sources, optical amplification, nonlinear management of systems, fiber media, etc., should be solved.

Challenge II What is the Shannon limit for single-mode fiber transmission systems? How can technology evolution go beyond the limit?

Given the constraint by the nonlinear impairment of single-mode fiber links, 100 Tbit/s is widely considered in the industry as the capacity limit of single-mode fiber transmission systems. According to Shannon's capacity theorem, the transmission capacity of optical fiber transmission systems is limited by physical characteristics including bandwidth resources and transmission impairments of the optoelectronic devices that make up the optical transmission link, and the correlation between the refractive index of the optical fiber and the power density of the optical signal results in a nonlinear variation of the frequency and phase of the optical signal with its power. In the transmission of optical pulse signals through a fiber channel, the pulse shape will be changed by CD, PMD, and the interaction with ASE, and nonlinear phase noise (NPN) will be caused by the fiber nonlinear effect, greatly affecting the phase modulated signal and it deteriorates with the increase of phase modulation level. In addition, stimulated Raman scattering (SRS) and stimulated Brillouin scattering (SBS) lead to the weakening of optical signals through energy transfer and result in noise interference. Limiting the transmitted optical power, destroying the phase matching condition, and the electric domain equalization algorithm are common strategies to suppress and compensate for nonlinear effects, but they make a limited contribution to improving transmission performance in practical applications. Using the neural network technology and information theory technique for fiber nonlinear compensation might be an effective means for silica single-mode fiber transmission systems to approach the nonlinear Shannon limit.

Challenge III What are the challenges facing broad-spectrum low-noise EDFAs in the extension trend of C+ L-band?

To increase the 400G PM-QPSK signal baud rate for long-distance transmission to 128 GBaud, the channel spacing needs to be broadened to 150 GHz and 80-wave multiplexing requires the C6T+L6T optical fiber spectrum. For broadening the L6T optical fiber spectrum, the longest wavelength approximates 1,627 nm. The commercial EDF previously used could support longest wavelength around 1,610 nm. How the wavelength of L-bands can be broadened to 1,627 nm is the major challenge facing the broad-spectrum EDFA. In the industry, a popular way to broaden the traditional EDFA to L6T broad spectrum is to raise the doping concentration of erbium particles and co-dope multiple elements such as cerium and phosphorus. Meanwhile, the overall erbium particle emissivity of erbium fibers on the L-band is several orders of magnitude lower than that on the C-band. Maintaining the population inversion level in longer erbium fibers requires multiple injections of high pump power, so as to output high-power amplified signals, and make the NF of the L-band EDFA significantly increase compared to that on the C-band. To compensate for the decrease in gain efficiency of L-band erbium fibers, higher pump power and longer erbium fibers are required, which will increase the EDFA size and cost.

Challenge IV Following the G.654.E optical fiber, which of the next-generation backbone fiber technologies will stand out?

The G.654.E optical fiber stands out with an ultra-low loss and a large effective area and wins over other fibers in high-speed transmission at and above 400 Gbit/s. It has found a wide application in new backbone optical cable network projects of operators. For the development of optical networks in the future, there are three development paths for optical fibers and cables: First, continue to reduce fiber loss and increase the effective area of single-mode fibers, but how a balance could be stricken between the effective area and the bending loss, thus making the fibers with increased effective areas more useful is a major problem; second, develop the SDM technology based on few-mode multi-core fibers. The two-dimensional uniform distribution of few-mode multi-core fibers in the axial and radial distributions over long distances, further reducing fiber loss, and the multi-core gain fiber technology are the challenges to be responded to for subsequent large-scale commercial application; third, hollow-core fibers are different from existing fibers in the materials technology and boast the advantages of ultra-low loss, ultra-low latency, low-nonlinearity, etc. They are in the lab research stage presently. The major challenge facing the application of hollow-core fibers is to improve their long-term reliability, attenuation stability, and robustness and redundancy of their manufacturing process.

Challenge V How will WSS evolve and develop as a core device of OXC in the future?

Currently, WSS devices supporting 32 dimensions bi-directionally have been widely applied to ROADM/OXC equipment. With the further increase in cross capacity and scheduling capability of optical network’s backbone nodes and the application of 400 Gbit/s single waves, new challenges are posed to the port number and integration level of WSS devices and the extension of C+L bands. In terms of the port number, WSS supporting 48 dimensions bi-directionally has been mature and is evolving to 64 dimensions. In terms of integration level, Twin WSS supporting bi-directional application has been widely used and is evolving toward Quad WSS supporting four directions. In terms of band extension, WSS devices supporting the C6T/L6T extension have been sold as products. The C+L integration will be realized in the future and there is a plan to further extend them to S-band. There are still many challenges in the development and evolution of WSS devices that call for industry-chain-based solutions, such as more simplified optical path design, reliability guarantee at high integration level, breakthroughs in materials (such as ultra-low-loss lens, optical grating, LCOS devices with super-large and super-fast deflection angles, and metasurface materials), and algorithms (compensation algorithm and control algorithm).

Challenge VI Is it possible to further extend bands such as S-band and E-band? What are the challenges?

Extending the spectrum of L, S, and E-band fiber channels beyond the C-band is a recognized means of significantly increasing single-fiber capacity. Extending from traditional C4T to C6T can increase the spectrum bandwidth capacity by 50%, and further extending to C6T+L6T can raise the spectrum bandwidth capacity by 200%. In addition, spectrum extension can also be conducted on S-band and even E-band. Yet, while optical amplification is carried out through an EDFA on C- and L-bands, on S- and E-bands, the doping of other elements is needed. Since spectrum extension will cause serious SRS effects, resulting in severe degradation of performance of shortwave channels, the realization and commercialization of future ultra-wide-spectrum optical transmission systems call for research and innovation in terms of the theoretical EGN modeling that describes SRS effects, the SRS compensation technique, the flexible spectral efficiency technology, new fiber links, etc., to overcome the impact of SRS effects on the performance of optical transmission systems in the case of ultra-wide spectrum extension.

Challenge VII In the general trend of energy conservation and carbon emission reduction, how can equipment and equipment room facilities evolve to meet the carbon peaking and carbon neutrality goals?

The boom of cloud computation and data-center business have driven the rapid development of optical network node transmission and switching capacity. Currently, ROADM/OXC at some core hub nodes has been upgraded to 32 dimensions, and the single-direction 80x400 Gbit/s demand has existed, bringing challenges to the relevant sites in terms of equipment space occupation and power supply heat dissipation. Using the photoelectric hybrid cross OTN + ROADM equipment at some co-station nodes of transmission networks and data center networks can efficiently meet the scheduling needs, and transforming the structure and power supply and heat dissipation modes of transmission equipment to match the racks being 800 mm deep, high-voltage DC power supply, the configuration of air going in from the front and going out from the back, and even liquid-cooling heat dissipation at data center rooms, will be an effective way to eliminate the conflict.

Challenge VIII How can in-depth integration of optical networks and digital twins be carried out effectively?

At present, the in-depth integration of the digital twin technology and optical networks has been a focus of research, which will give rise to the ChatGPT-based intelligent application of optical communication. Optical networks are developing towards automated and intelligent operation in the whole lifecycle of “planning, construction, maintenance, and optimization”, which can effectively enhance the network performance and resource utilization rate. Yet, there are harsher requirements to be met in practice. To build digital twin networks with operation mechanisms, behavioral rules, and health status highly consistent with those of physical optical networks, the inconsistency between digital twin networks and physical networks caused by the static mechanism modeling and dynamic variable light factors of optical networks should be tackled first. For the self-intelligence of L4/L5 high-dimensional networks, new generalized mechanisms should be come up with to deal with the problem that optical network intelligent models can’t adapt to multiple O&M scenarios, and efforts should also be made to optimize and upgrade technologies for intelligent models in complicated scenarios such as different velocities, modulation modes, and fiber types, to build automated intelligent platforms comparable to human experts in operation and maintenance capacity. In addition, for the current cross-region vendor optical network systems, effective cross-region collaboration mechanisms and standards should be set, to unify software, hardware, and protocols and exchange data, so as to ensure the efficiency of cross-region collaboration and support traditional networks to smoothly evolve toward intelligent optical networks and self-intelligent networks.

Challenge IX As the core of optical communication systems, how can high-speed optoelectronic devices further break bandwidth limitations?

High-speed and high-bandwidth photoelectric devices (above 400 Gbit/s and 130 GHz) are necessary for the leapfrog development of optical development. Take high-speed modulators as an example. Currently, industry players are focusing on the III/V photonic integration represented by silicon optical integration and InP, and the technology roadmap of three materials (lithium niobate films, etc.) of the thin film material system. The characteristic frequency of InP is about 160 GHz, and with the design and optimization of the process and structure, it is expected to support the high-bandwidth 400G and above long-distance application of modules and systems. Traditional silicon optical modulators require a high output amplitude of drivers, and the theoretical bandwidth limit is about 90 GHz, which is difficult to realize in 400G and above applications. The thin film material system (lithium niobate films, etc.) has an ultra-high theoretical bandwidth (> 200 GHz) and a low loss, but faces certain challenges in functional integration (detectors, VOA, polarization control, etc.). Hopefully, combining and integrating silicon light and thin film materials may create competitiveness in integration level and high bandwidth.

Challenge X As optical network chips get increasingly complex, will the existing technologies become obsolescent or enter a new development stage?

Since optical networks need to develop toward super-large capacity, super-long distance, super-high speed, and extremely high intelligence, it is necessary to produce larger-capacity, higher-speed, lower-power, and highly reliable chips. Efforts should be made to enhance the integration level and reduce the power consumption of the chips through the continuous evolution of their production processes. However, merely increasing the number of transistors per unit area, reducing power consumption, and improving the logical speed and performance through the continuous evolution of production processes is increasingly difficult to meet the needs. The balanced application of advanced chiplet packaging and production processes might be an effective way to develop chips for future optical networks. While traditional electric chips are in an evolution dilemma under Moore's Law, optical modules can eliminate discontinuity points of critical impedance through the photoelectric sealing of DSPs, modulators, drivers, receivers, etc., on common substrates, thus significantly reducing reflection and improving bandwidth. Photoelectric sealing is key to developing high-baud-rate and high-bandwidth optoelectronic devices in the future.

Customer Points of View