Trend 1: Full Opticalization of Networks
From the demand side, Wei Leping proposed that microprocessors have developed from a single core to thousands of cores of Tera-level computing; supercomputer capabilities have increased thousands of times in ten years, and it is expected to reach 100 billion billion times per second in 2025; video will become the first driver power, the traffic is close to 2/3 of the network, AR/VR will increase the capacity demand; the super perception and response of high-end IoT machines requires higher-speed bandwidth and low-latency connections; in addition, other new application requirements, such as low Latency/jitter, determinism, high availability, etc.
On the supply side, the fiberization of transmission links is approaching 100%, and the fiberization of access networks has reached 93%, marking the end of all-optical transmission and access on the network side (all-optical network 1.0 stage) . The photochemistry of the network trunk transmission and switching nodes is about to be completed, and it is expanding to the metropolitan area access network. In general, the actinization of the entire network is moving from the stage 1.0 to a new stage of true all-opticalization in 2.0!
Trend 2: High capacity of all-optical network transmission links
Wei Leping mainly introduced from the two directions of DWDM and TDM. Among them, the main direction of DWDM is that the traditional C-band 80 wave can be extended to C-band 96 wave and extended C+-band 120 wave with a small cost and technical transformation, and the expansion gain of 20% and 50% can be obtained respectively. At present, the latest trend is to expand the C+ band 120 waves plus the L+ band 120 waves for a total of 240 waves, and the expansion gain is expected to be as high as 200%. The main challenge is to balance the Nyquist filter compensation and amplifier performance.
In the TDM direction, mainly using the new oDSP, the single-wave 400Gbps transmission distance of QPSK based on 130G baud can be expanded from 600 kilometers to 1500 kilometers (after 2023), which can cover 99% of the trunk line multiplexing section distance.
Trend 3: High capacity of all-optical network switching nodes
Wei Leping said that the expansion trend based on wavelength switching is currently dominated by 20 dimensions. The 300T of the 32-dimensional ROADM can meet the current demand for the largest node capacity. The 600T of the 64-dimensional ROADM can meet the demand for the largest node capacity in 2023. The multi-fiber space division multiplexing and switching based on traditional physical isolation has low blocking rate, slow growth, good optical transparency and great capacity expansion potential. Therefore, in the short and medium term, node capacity can continue to rely on ROADM capacity expansion by wavelength switching; in the medium and long term, nodes and links will have to rely on multi-fiber space division multiplexing and switching technology.
Trend 4: Continuous optimization of all-optical network recovery time
It mainly includes optimization at the hardware and software levels. At the hardware level, Wei Leping said that the typical WSS switching time is about 1 second, and there is little room for improvement; the key to the OTU switching time is the switching of the laser wavelength, and some laboratories have been able to reduce the OTU switching time to 1 second through control and algorithm optimization. within 3 seconds.
At the software level, mainly by introducing "centralized routing calculation + distributed control" to replace "distributed calculation + distributed control", it can avoid the conflict of wavelength, relay and routing, and reduce the recovery time. Through the network-wide topology abstraction of PCE and SDN, failure recovery pre-computation can be performed by using CPU idle time, thereby reducing the calculation time of route recovery. Machine learning is introduced to predict optical performance degradation, optical fiber or equipment failures, save service commissioning and recovery time, and even implement active rerouting, greatly reducing recovery time.
Trend 5: Cloudification of all-optical networks
IDC predicts that in 2025, more than 90% of applications in China will be migrated to the cloud, and DC will be fully cloud-based. As the network supporting the application, realizing that the network moves with the cloud is the biggest driving force of cloudification. Except for high real-time, high sensitivity and local applications, all areas of the network will be fully cloudized.
In addition, the traditional closed and rigid network itself is developing from a hardware-based architecture to a deep transformation of software, virtualization, cloud, intelligence, and service, and all-optical networks are no exception.
It is worth noting that, through the introduction of SDN, the first realization of the software of the all-optical network is the premise of cloudification. Because SDN means the decoupling of software and hardware of the all-optical network, the connection and function will only be flexibly determined by the software, which will facilitate the subsequent evolution to cloudification, intelligence, and service, and realize rapid automation and intelligence of networks and services. deployment and continuous evolution, upgrading and innovation.
Trend 6: Intelligentization of all-optical networks
Wei Leping pointed out that the implementation of centralized management and control of SDN can greatly improve the efficiency of operation and maintenance, but the establishment/removal of optical paths must rely on manual instructions, and it is difficult to achieve active network reconstruction and active operation and maintenance.
In the performance of all-optical network intelligence, Cognitive Optical Network (CON) is one of the typical ones. This is a new generation of intelligent optical network based on machine learning, which can automatically perceive, understand and learn the external environment, and adjust in real time. Network configuration, intelligently adapt to changes in the external environment. At its core is a cognitive decision-making system that manages transport requests and network events. The control and management system is responsible for controlling and disseminating the relevant signaling. It can not only automatically optimize optical network configuration, but also quickly detect and locate faults, monitor real-time optical path performance and predict quality, automatically optimize transmission parameters, implement traffic forecasting and routing planning, perform fault root-finding, and reduce optical layer recovery time. The overall quality of the all-optical network.
Trend 7: Openness of all-optical networks
In order to cope with the severe situation of weak industry development, the development experience of the IT industry and the opportunity of introducing SDN/NFV/Cloud are used to realize the decoupling of inter- and intra-layer functions, reduce costs, and create an open industrial ecology to become a sustainable telecom industry. The key to development and consensus. According to Wei Leping, SDN means decoupling of software and hardware and software-based network functions, which is the basis for network opening. In addition, starting from the wireless access network, various fields of the network are gradually opening up, such as interface standardization, software and hardware decoupling, opto-decoupling, hardware white boxing, software open source, etc. All-optical networks are no exception. It is one of the fastest moving areas. Wei Leping also mentioned that the opening steps mainly include opening optical line systems, opening optical switching nodes, and opening functional blocks.
Trend 8: Ubiquitous all-optical network
With the continuous development of demand-side applications and the continuous reduction of supply-side equipment costs, the all-optical network is beginning to expand to the network edge, moving towards an end-to-end ubiquitous all-optical network. Wei Leping mentioned that both the network transmission side and the network access side are changing. He proposed that the long-term goal of the all-optical network is to become a ubiquitous optical socket like an electrical socket.
Trend 9: Optimization of all-optical network cost
On the network transmission side, the key is technological innovation and economies of scale. Innovation at the physical layer is to remove unnecessary functions at the edge of the network and relax unnecessary harsh temperature requirements; develop a new generation of optical switching devices. At the network layer, it is a "gray box" or even a "white box" system controlled by SDN, software and hardware decoupling, and optoelectronic decoupling, which promotes the opening and prosperity of the all-optical network ecosystem. In terms of architecture, a new metropolitan area network architecture with converged bearer should be introduced in combination with the deployment of edge cloud. At the same time, it is also necessary to realize the ITization of edge DCI and other equipment, including open architecture, interface standards, software and hardware decoupling, opto-decoupling, protocol reduction, software open source, gray box/white box, manageable and controllable, etc.
On the network access side, the key is still technological innovation and economies of scale. Similar thinking and different specific innovative technologies, highly sensitive cost is the challenge. Finally, it is necessary to standardize the unified F5G.
Trend 10: Coordinated development of all-optical access and 5G/6G
The all-optical network is not only the best bearer of 5G/6G, and its optical access segment is also a competitor of 5G/6G. The two can only coordinate and synergize, and each has its own strengths and cannot be neglected.
Wei Leping explained it in detail from the following aspects. In terms of business applications, 5G/6G focuses on data services and short videos with medium and small screens, medium bandwidth and quality, and data services and videos with large screens, high bandwidth and high quality on the optical access side. In terms of business models, optical access is not sensitive to traffic and usually adopts a monthly subscription system, while 5G/6G is sensitive to traffic and focuses on the tiered traffic system with limited traffic. 5G focuses on the speed below 50Mb/s, which is more economical. The Gigabit optical access network is not sensitive to the speed, and focuses on the speed above 50Mb/s. Fixed-Mobile Convergence will gradually move from the traditional unsuccessful Fixed-Mobile Convergence (FMC) to a new stage of Wired Wireless Convergence (WWC) under the 5GC single-stack protocol. Industrial Internet scenarios, the two should focus on mobile and fixed scenarios respectively
