Recently, at the main forum of the NGON/DCI Summit in Nice, France, Wei Xueqin, General Manager of Fiberhome R&D Department, delivered a keynote speech entitled “AI-based Intelligent Optical Network”, which analyzed the evolution trend of next-generation optical networks and introduced FiberHome intelligent optical network solution.
Traditional optical network faces bottlenecks in development
Traditional optical networks face three major pain points. First, network planning, operation and maintenance, and capacity expansion is inefficient. Manually configuring services makes it difficult to adjust the network. Secondly, optical network optimization is time-consuming and laborious, and the network optimization process is prone to errors, and it is difficult to achieve satisfactory optimization results. Finally, network fault diagnosis and positioning is difficult, and many complex network problems often require experts to participate in troubleshooting.
AI can reduce network CAPEX and OPEX
AI technology is an important method for the above three types of pain points. Based on AI, 80% of manual operation and maintenance operations are converted to machine processing, which helps operators reduce costs, improve user experience, and reduce optimization time.
In the artificial intelligence optical network, there are four typical application scenarios.
The first application scenario uses artificial intelligence technology to implement optical network fault diagnosis. For example, through training, the machine gains intelligence, which can automatically and quickly find out which optical module is causing the network to be unreachable, or to determine which segment of the fiber has a problem. The intelligent optical network diagnosis scenario can shorten the troubleshooting period to less than 8%.
The second scenario, artificial intelligence technology is used to predict optical network faults. Fault prediction can help administrators prevent optical network failures in advance. Based on advanced algorithms, the accuracy of fault prediction can be increased to 97%.
The third scenario, based on artificial intelligence technology, we can reduce the optical network performance evaluation error rate to less than 2%.
The last scenario, based on artificial intelligence technology, we can achieve network intelligence optimization, and the efficiency is improved by six times.
The problem that the next generation optical network needs to solve when introducing AI
Applying artificial intelligence technology to next-generation optical networks can bring many benefits. But how do you apply AI in NGON? What are the most critical elements in the application process?
First, there is a need to address the challenge of data collection. Effective data cleaning, data annotation, and data analysis methods can help us achieve AI applications in NGON faster.
Secondly, AI-based next-generation optical networks require a large number of algorithm models for traffic prediction, network optimization, and fault location. For each application scenario, you need to design a specific algorithm.
Finally, in order to support real-time intelligent reasoning, it is necessary to increase the computing power and enhance the computing power with a high-performance infrastructure.
FiberHome intelligent optical network solution accelerates smooth network evolution in the future
FiberHome artificial intelligence-based 5G bearer solution perfectly supports the smooth evolution of next-generation optical networks, consisting of two platforms and one network.
C-ONE platform: The letter C indicates cooperation, focus, and cloud, and ONE indicates open network engine. The C-ONE platform enables unified management and control of optical networks. This platform utilizes the model algorithm generated by the artificial intelligence training platform to automate the processing of FCAPS (network management) and COMPA (SDN) tasks to realize optical network planning, installation, service analysis, fault diagnosis, and capacity expansion.
Artificial Intelligence Training Platform: Generating a series of intelligent models using offline models. These intelligent models are used for network planning, traffic prediction, and performance optimization. AI-enabled optical network equipment can more accurately collect network health data and more flexibly implement various strategies issued by C-ONE. Optical network equipment has simple basic intelligence to clean network data.
IP+ optical infrastructure: The network device uploads the collected data to the C-ONE platform through telemetry or sFTP, and distributes it in the C-ONE. Deploy management and control services on a unified micro-service platform. The four major types of AI engines, namely, the analytics engine, the forecasting engine, the intent engine, and the automation engine, provide intelligent services for upper-layer applications.
Wei Xueqin concluded by summing up: "Intelligent optical network is the trend of development. In the future, the traditional management and control platform will evolve into the C-ONE platform. With AI, C-ONE will become the brain of the future optical network. FiberHome is willing to work with operators and industry colleagues to promote smart optical network commercialization, to meet a brighter future."
From enterprise |
.jpg)
