[Shenzhen, China, June 26, 2018] Today, Huawei announced the launch of its optical service fault prediction solution based on artificial intelligence (AI) and Big Data technologies. This solution enables carriers to predict faults before they occur on their optical networks. Carriers can then preemptively implement traffic grooming and troubleshooting according to the predicted fault types, comprehensively improving the customer experience.
As the cornerstone of our information society, optical networks are indispensable to the Internet, 4G/5G, electrical, financial, and other industrial networks. This is evident in the fact that if optical networks become faulty, end users may experience service interruption throughout the time-consuming troubleshooting process. For this reason, carriers demand a solution to detect optical network risks before they become faults, delivering "zero fault" user experience.
Statistics show that 60% of optical service faults are associated with optical channel deterioration, optical module deterioration, or bit error rate (BER). Traditional O&M relies on setting empirical thresholds, which is insufficient for preemptive fault detection. Using traditional methods, even after faults occur, carriers cannot immediately identify the fault types or quickly resolve issues. This is primarily because the fault data is not centralized and cannot be associated. Ultimately, this leads to customer complaints.
Hardware and service faults usually present early symptoms, such as being unstable or experiencing performance deterioration. Unfortunately, traditional O&M is generally unable to discern such signals. In response to such challenges, Huawei makes a huge breakthrough by applying Big Data and AI technologies to the predictive maintenance field. The former can collect key performance indicators (KPIs), and the latter, through trend learning, model training, and model deployment, can accurately detect what has happened, what will happen, and why.
Optical service fault prediction is an application of Huawei's Intent-Driven Network (IDN) architecture. It combines Huawei Network Cloud Engine (NCE) and optical devices to collect real-time BER, optical power, and other KPIs of optical services by using Big Data technologies. It then predicts and scores a wide range of possible service risks (such as optical channel or multi-segment service deterioration). More importantly, it provides causes and suggestions for mitigating these risks.
Preliminary testing on carrier OTN networks verifies that this solution can predict optical channel and multi-segment service deterioration, helping the carrier improve network robustness and significantly reduce preventive maintenance inspection (PMI) costs. On the live network, the solution has already achieved a prediction accuracy of 85%, and this will further increase with continuous self-learning and self-training of AI.
Lu Hongju, General Manager of Huawei’s NCE Domain, said: "As an application of Huawei IDN, the optical service fault prediction solution can be applied to Huawei's optical network solutions. Its core AI technologies, after training and optimization, can predict optical service faults on various devices, such as routers and switches. In the future, more AI-aided applications will be developed to solve the problems that traditional technologies cannot. With key algorithm breakthroughs and various training models, we believe that IDN will reduce carriers' workload during device maintenance and allow them to focus on providing users with a wider range of designs and better services."
Huawei will help carriers and enterprise customers transform their networks from device-centric to user-centric. Cloud, Big Data, and AI technologies can be applied to build intelligent, simplified, ultra-broadband, open, and secure IDNs, enabling carriers and enterprise customers to achieve IDN automation and intelligence. The ultimate goal is to bring digital to every person, home and organization for a fully connected, intelligent world.
If you are interested in Huawei's optical service fault prediction solution, we welcome you to visit us at the Next Generation Optical Networking (NGON) Conference in Nice from June 26 to 28, or Mobile World Congress Shanghai from June 27 to 29.