Nine Key Challenges Facing Optical Communications in the Next 10 Years
Our predictions for the challenges and solutions required to take optical communications to the next level.Huawei Tech Issue 93
Ever since Nobel Prize winner Charles Kuen Kao presented his optical fiber research to the world in 1966, optical fiber has shown great potential and will continue to do so – optical technology underpins the information age and the global economy. Based on more than 25 years of expertise in optical communications, we’ve identified nine potential technological challenges facing optical communications in the next decade.
More than 25% annual growth in data services over the last 20 years has resulted in long-distance systems characterized by the following features:
First, the transmission distance of a long-distance network is at least 1,200 km.
Second, long-distance systems use C-band for spectral capacity, with wavelength path spacing a multiple of 50 GHz or 25 GHz.
Third, the span of a long-distance system is about 80 km.
To further develop long-distance systems, we can consider the following factors:
400 Gbit/s x 80-wavelength optical coherent system: This includes innovations like high-performance 400 Gbit/s optical modules, C-band and L-band frequencies, a wideband steady-state optical system to offset non-linear effects, and automated network O&M to stimulate the large-scale commercial adoption of 400 Gbit/s systems.
Second, single-fiber capacity of 100 Tbit/s (1.2 Tbit/s x 80 wavelengths): Based on the existing G.65x fiber channels, research should be conducted in the three areas: raising the spectral efficiency (SE) stretch target from 2.67 (corresponding to 125 GHz for 400 Gbit/s) to 4 (corresponding to 300 GHz for 1.2 Tbit/s), raising the available frequency band stretch target by 150% from Super C+L (10 THz) to C+L+S+U/E (> 24 THz), and improving the signal-to-noise ratio.
The key index for measuring the carrying capacity of an optical network system is the transmission capacity of optical fiber. However, improving the single-fiber transmission rate in an existing optical fiber transmission network presents a major technical challenge.
As transmission algorithms evolve, engineering capacity is approaching its theoretical limit, and the manufacturing of new high-frequency components is becoming increasingly difficult, indicating that greater challenges are ahead for single-wavelength acceleration technology. Exploring new available frequency spectrum for fiber transmission systems is the way to uncover new innovations for expanding transmission capacity in the optical network industry.
To develop new-band fiber communication systems, it’s important to develop fiber amplifiers that support new spectral applications. Research in optical amplification technologies based on C-band and L-band has made substantial progress, but technologies based on S-band (1460–1530 nm) are still in development. Key technologies such as rare-earth-doped (Thulium and Bismuth) gain fibers and semiconductor optical amplifiers (SOAs) are important for amplifying and transmitting S-band optical signals.
Currently, single-mode optical fibers such as G.652, G.655, G.653, and G.654 are used for WDM system transmission. These fiber types have the following weaknesses:
For G.652 fiber, loss and nonlinearity are key constraints affecting coherent transmission.
The transmission distance of G.655 fiber is 40% less than G.652 because of strong nonlinearity caused by low dispersion and the small effective cross-sectional area in the coherent era.
For G.653 fiber, four-wave mixing causes serious nonlinear interference between wave channels in a DWDM system. This results in low incident optical power that is averse to multi-channel WDM transmission over 2.5 Gbit/s.
For G.654 fibers, higher-order-mode multipath interference (MPI) greatly impacts system transmission and cannot meet transmission requirements for expansion to S, E, and O bands.
To exponentially increase capacity at unchanged distances and comply with the optical Moore's law, next-generation optical fiber must have the following characteristics:
First, high performance, low intrinsic loss (<0.14 dB/km), and strong anti-nonlinear effect ability; second, large capacity, covering all or wider available bands (e.g., C, L, S, E, and O bands); and third, low cost plus support for engineering approaches such as ease of manufacturing with costs comparable to G.652 fiber, and ease of deployment and maintenance (fiber routing and splicing).
Future research should include hollow-core optical fibers, SDM optical fibers, and more.
In a WDM transmission system, the effective area of an optical fiber is less than 80 μm2. As a result, even a relatively low incident optical signal power will generate nonlinear effects such as distortion between optical signals and physical channels, and between different signal channels in an optical fiber. Different types of effects are generated depending on the mechanism: stimulated scattering effect (stimulated Brillouin scattering (SBS) and stimulated Raman scattering (SRS)), the Kerr optical effect (self-phase modulation (SPM)), cross-phase modulation (XPM), four-wave mixing (FWM)), and more.
Currently, optical transmission systems are evolving towards single-fiber 400 Gbit/s x 80 wavelengths and higher capacities. As the transmission rate and device bandwidth increase, signals become more sensitive to nonlinear distortion. However, optical systems now occupy wider frequency bands (for example, C and L bands), which means that their total incident optical power is higher than that of a C-band optical system. This results in a stronger signal nonlinear distortion effect. Therefore, nonlinear channel compensation algorithms are key to improving the capacity of next-generation optical transmission systems.
Research and development on nonlinear channel compensation algorithms should focus on algorithm complexity as well as compensation. This can help achieve targets for low chip resource/power consumption.
Current methods of overcoming nonlinear signal distortion in an optical channel include a theoretical nonlinear model that resembles the actual channels and an accurate and concise nonlinear compensation algorithm. These methods are key research areas for improving optical fiber capacity.
Fiber broadband networks are widely used in fields like home Internet access, enterprise operations, government services, and transportation management. There will be a sharp increase in optical network nodes, heralding an era of ultra-large-scale optical networks.
A typical problem in optical network planning is routing and wavelength assignment (RWA), which has proved NP-hard, where the wavelength assignment (WA) subproblem is equivalent to the graph coloring problem. As network scale increases, it is exponentially more difficult to find the optimal solution. The growth in emerging data service applications leads to an increase in the scale of networks. As a result, network planning has become a more complex and diverse problem than RWA. For example, one or more protection routes needs to be planned for different fault scenarios, taking into consideration a range of new network planning problems. These include mapping between pipes of different levels and sizes, relay minimization, and topology optimization for network expansion. In upcoming ultra-large-scale networks, there will be thousands of nodes and tens of thousands of services. Planning an optical network of this scale will undoubtedly be more challenging.
The explosive growth of data traffic poses great challenges on the processing capacity and scheduling capability of backbone nodes on transport networks. A wavelength selective switch (WSS) supports large-granularity service grooming and ultra-low latency. It is a core functional module of ROADM/OXC and is also the ideal component for the traffic surges and ultra-low latency requirements of future optical networks.
With higher requirements on network bandwidth, network latency, and flexible scheduling of network services, WSS modules are evolving towards more ports, faster scheduling, and higher performance.
Port numbers: Mesh networks require higher-dimensional service-grooming capabilities. Next-generation 128D + WSS is expected to become a reality based on breakthroughs in materials such as LCoS supporting deflection angles ≥ 11°.
Ultra-high optical performance: Future-proof high-performance WSS solutions (optical performance: insertion loss (IL) ≤ 3 dB, isolation ≥ 35 dB, frequency offset: +/-0.5 GHz, polarization-dependent loss (PDL) ≤ 0.3 dB, wider bands supported) are expected with breakthroughs in design and materials.
Ultra-low latency: The μs-level WSS switching rate is expected to be achieved through breakthroughs in design, materials, and algorithms.
How can we develop a high-performance, high-dimensional, and highly reliable WSS solution that meets the requirements for next-generation transmission? We believe that areas of research should include: simpler optical path design, material breakthroughs (such as ultra-low loss lenses, gratings, ultra-large-angle and ultra-fast deflection LCoS, and super-surface materials) and algorithms (including compensation algorithms and control algorithms).
As network bandwidth evolves (10 Gbit/s →50 Gbit/s→200 Gbit/s), the emergence of new services (AR, VR, and holographic) and applications (manufacturing and wireless bearer) requires more in terms of latency, jitter, and the security isolation of optical networks.
The evolution of the PON technology has two constraints: 1. Evolution must be based on the deployed ODN. 2. The energy consumption per bit and cost of the next-gen technology must be at least half of previous generation’s.
The challenges are as follows:
First, the transmit optical power of the transmitter is beyond current technology. Depending on the ODN (multi-level optical splitting, 20 km) deployed on the live network, the ODN power budget should generally be larger than 32 dB. The transmit power required for 200 Gbit/s bandwidth is about 30 dB (17 + 13). Currently, the required transmit optical power of the transmitter has exceeded what's technically viable (see Figure 1).
Figure 1: Receiver sensitivity vs. bandwidth (Source: ITU-R SM.575-2 Recommendation)
Two viable options for achieving a breakthrough are with high-bandwidth, high-power, low-chirp transmitters and new modulation and demodulation technology. Other viable options are yet to be explored.
Second, existing PON architecture cannot meet the needs of service growth. To meet network requirements for ultra-low deterministic latency, jitter, and hard isolation in target scenarios, a conventional TDM PON mechanism needs to use multi-ONT uplink framing. This will result in DBA algorithm scheduling latency and bandwidth overhead for burst alignment of different ONT frames (the lower the latency/jitter, the higher the bandwidth overhead). Therefore, the new architecture requires new optical systems, components, and algorithms, as well as modulation and demodulation mechanisms, to achieve collaborated breakthroughs.
Network connectivity is the foundation of Internet communications, and satellite Internet is no exception. Laser technology is used for inter-satellite communications, because it has the advantages of small divergence angle, large transmission capacity, long transmission distance, and resistance to interference/interception. However, there are still many technical challenges to overcome to build inter-satellite optical communication systems suitable for commercially viable IoT satellite constellations with low Earth orbit (LEO) satellite networks at a large scale.
Most DWDM optical modules today output one channel of signals at one wavelength (very few vendors offer two-wavelength, two-channel products). Each wavelength requires an independent laser, modulator and control circuit, digital signal processor (DSP), and clock, power supply, and central control on the optical module. Multi-wavelength multiplexing is implemented outside the module, and extra slots are required for multiplexer/demultiplexer boards, which occupy a lot of space in equipment rooms. As network traffic increases, we will see the commercial use of L, S, and U bands, which will require even more equipment room space.
Therefore, we believe there needs to be a single optical module that can output signals at multiple wavelengths. It needs to be able to integrate more than 100 wavelengths in all bands (C, L, S, U, etc.) to implement multiplexing and demultiplexing inside the module, so that only one module is needed for each fiber and each slot. This component model will inevitably bring great technical challenges. We believe that research should focus on optical frequency comb technology, hetero-integration technology, photo-electronic co-packaging technology, and heat dissipation technology.
In-depth thinking and action are both needed for breakthroughs in key optical communications technologies. Huawei is ready to collaborate with upstream and downstream players in the value chain of the optical communications industry and top optical communications experts and scholars around the world to overcome challenges in optical systems, optical components, optical algorithms, and optical intelligence.