With the implementation and promotion of the "Broadband China" strategy, broadband users in China are developing rapidly, and the broadband popularity is soaring. The number of internet broadband access users has exceeded 480 million, of which the number of fiber-to-the-home/office (FTTH/O) users has exceeded 448 million, accounting for 93.3% of the total internet broadband access users. In addition, increasingly diverse home services such as high-definition videos, VR/AR and cloud games make home broadband users pay more attention to service experience, thus posing higher requirements on broadband network quality.
As the number of optical broadband users grows at an explosive rate, weak optical power problems on fiber broadband links have become increasingly prominent. According to the analysis of user complaints of an operator in China, the proportion of user complaints caused by weak optical power on fiber broadband links reaches 36%, so weak broadband optical power is an important factor that affects service experience of the broadband users and may even cause service failure.
Traditionally, weak optical power faults are identified and located manually. Due to the large number of optical broadband users, the problems of weak optical power are quite common. The diversity of topology distribution also makes rectifying weak optical power difficult, time-consuming and inefficient. Moreover, rectifying weak optical power on ODN fiber links involves OLT PON interface, backbone optical fiber, level-1 splitter, branch optical fiber, level-2 splitter, to-home optical fiber and ONU. It takes a lot of time and manpower to sort out the faults one by one. Since it is impossible to determine the unreasonable networking mode of multi-level splitters, it is also very difficult to check the weak optical power faults manually.
ZTE's AI-based precision location solution for weak broadband optical power is an automatic fault location system using big data and AI analysis based on Athena 2.0. The system automatically, frequently and fully collects various optical link feature data of OLT and ONU in the optical broadband access network. After denoising, converting and analyzing the optical link feature data, it uses AI clustering algorithms for intelligent analysis, compares and learns based on the background knowledge base and optical link fault model base, and finally accurately identifies the location and cause of weak optical power links. The solution can guide O&M personnel to actively and efficiently rectify weak optical power links, implement network O&M, and reduce the number of fault reports.
The Athena 2.0-based precision location system for weak broadband optical power is easy to deploy and use (Fig. 1). It is deployed on a PC server and uses the TL1 interface to connect EMS for automatic synchronization of resource data in the optical broadband access network. It uses the SNMP protocol to directly access OLT to collect optical link data of the entire network, and automatically locates and analyzes the cause of weak optical power based on AI. The O&M personnel can log in to the system through WEB, obtain an report on rectifying weak optical power links, and complete the fault rectification with ease according to the report.
Automatic, Frequent and Full Optical Link Data Collection
The Athena 2.0-based precision location system for weak broadband optical power automatically collects dozens of optical link feature data of OLT and ONU in the optical broadband access network, including optical module type, optical layer alarm, OLT receive/transmit power, ONU receive/transmit power, packet loss rate, bit error rate and optical distance. The system also adopts multiple policies such as incremental collection and differential avoidance to improve the collection rate of weak optical power. The actual collection rate of weak optical power verified in the existing network reaches 95%.
Weak Optical Power Delimitation and Location by Big Data and AI
The PON access network divides ONUs in the same PON into different groups due to level-1 and level-2 optical splitting. These groups have certain cohesion and regularity in terms of optical power and distance. However, these features are dynamic rather than static, and their thresholds and rules identified manually cannot fully meet and cover all dynamic data distributions. It is therefore necessary to use AI-related algorithms to implement weak optical power detection. The AI-based weak optical power detection mainly uses the dynamic K clustering algorithm and the DBS fault detection algorithm to find the best clustering parameters in accordance with the constraints of optical splitting scenarios, and the contour system to dynamically evaluate the clustering effect. Based on the clustering result and distribution, the AI algorithms delimit and locate the causes of weak optical power, which involves determining fiber backbone, branch and to-home link failure, or optical split ratio over threshold. The actual accuracy of weak optical power fault delimitation and location verified in the existing network is up to 95%.
Visual Display of Weak Optical Power Causes
The Athena 2.0-based precision location system for weak broadband optical power automatically displays in charts the weak optical links analyzed by AI as well as their delimitation and location causes. The O&M personnel can access the system through WEB, obtain the data and causes of weak optical links, and efficiently complete the rectification task. As verified in the existing network, the rectification efficiency is increased by more than five times.
The system can also be interconnected with the work order system of a network operator, so that the weak optical link list can be automatically sent to the work order system for rectification. The work order of weak optical power rectification is linked with the weak optical power diagnosis of the precision location system, which ensures a closed-loop workflow. The system provides a variety of statistical charts for the O&M personnel to learn about the progress and effect of weak optical link rectification.
After the Athena 2.0-based precision location system for weak broadband optical power was deployed in three cities of an operator in China, it helped the O&M personnel accurately locate the causes of weak optical power, reducing the weak optical power ratios in the three cities from 6.14%, 10.49% and 9.01% to 3.11%, 3.42% and 3.48% respectively in a short period of time.
The system has been commercially deployed by Chinese operators in Guizhou, Yunnan and Zhejiang and will be widely promoted throughout the country. It helps operators rapidly identify weak broadband optical links, precisely locate the causes, and substantially improve the efficiency of weak optical power rectification. In this way, operators can actively eliminate the potential quality problems of their optical broadband networks, thereby enhancing user experience.