Network alarm analysis faces great challenges
With the expansion of the network and the increase of communication equipment, the number of network alarms has also increased dramatically. "At present, the number of IPRAN alarms in a province has exceeded 1.9 million, and more than 60% are emergency alarms. The traditional alarm processing mode is time-consuming and laborious, and the effect is not obvious. " China Unicom Network Research Institute expert said, " Mining internal associations from massive network alarm data, classifying derived alarms and root cause alarms, and accurately locating network equipment faults are huge challenges for communication network alarm analysis. " The traditional alarm processing method is based on the experience of network operation and maintenance experts. The alarm association rule is built to implement alarm filtering and alarm compression. The association rule often does not match the actual condition of the network. It is hard to accurately locate root alarms and fault points. "Traditional rule is essentially a static model. It cannot dynamically adjust analysis methods and strategies according to network alarm events, and accurately implement alarm correlation." Introduction of FiberHome's artificial intelligence experts, "A limited number of association rules cannot comprehensively analyze massive and complex network Alarms, unable to handle uncertain and bursting alarms in the communication network, and unable to intelligently sense anomalies in the network."
Using new technology and new methods to correlate and analyze network alarm events, find root cause alarms, accurately locate fault points, and solve the difficulties of current network alarm is huge challenge for operators.
Significant opportunities for network alarms artificial intelligence analysis
In response to the huge challenge of communication network alarm analysis, FiberHome and China Unicom Network Research Institute jointly launched a network alarms research project based on artificial intelligence, using artificial intelligence convolutional neural network and cyclic neural network model, data mining association analysis algorithm, deep exploring network alarm internal correlation rules, accurately locating root cause alarms, achieving accurate, dynamic, automatic network alarm correlation, alarm compression, fault location, and ultimately building a stable, reliable, intelligent communication network.
Through in-depth analysis of massive network alarm samples, FiberHome has mastered various types of artificial intelligence and data mining core technologies such as network alarm statistical analysis, regression prediction, and supervised learning. FiberHome has the theoretical basis and practical experience of building a machine learning model, dynamically and intelligently exploring network alarms through neural networks. Under the framework of Fiberhome and China Unicom's strategic cooperation agreement, the two sides jointly built a network alarm artificial intelligence analysis engine, integrated human intelligence and machine intelligence, and achieved intelligent analysis of network alarm events.
FiberHome and China Unicom jointly achieved the key technological breakthroughs of network artificial intelligence
In 2018, Fiberhome and China Unicom Network Research Institute strategically cooperated to jointly tackle the key technologies of network artificial intelligence and develop a network alarm intelligent correlation analysis platform. Aiming at a series of challenges faced by artificial intelligence technology in the communication network landing, FiberHome and China Unicom jointly set up a high-level expert team, starting from five aspects: training sample acquisition, intelligent platform construction, intelligent model tuning, application scenario screening, and analysis effect evaluation, conducting in-depth basic theoretical analysis, comprehensive technical program evaluation, and rigorous experimental comparative analysis. Through the joint efforts of technical experts and business experts, in the aspect of core technologies such as training sample cleaning, intelligent correlation analysis, accurate fault location, and automatic feedback evaluation, FiberHome and China Unicom have jointly achieved a series of major breakthroughs, which laid solid foundations for artificial intelligence technologies in communication networks.
Plug the smart wings into the communication network, let artificial intelligence help the optical network agile service, FiberHome will continue to work together with China Unicom to build a new generation of intelligent communication network.
From enterprise |
.jpg)
