In 2019, TM Forum proposed the concept of “autonomous networks (AN)”, which aims to provide innovative ICT services with Zero-X (zero-wait, zero-touch, zero-trouble) experience based on fully automated life-cycle operations of self-configuration, self-healing and self-optimization, empowering digital transformation of operators. Over more than two years, the industry has reached a consensus on the concept with the launch of multiple relevant standard or research projects. In September 2021, the TM Forum White Paper on Self-Intelligent Network 3.0 was released, which advocates “building an automated and intelligent network”. ZTE has also come up with its IP autonomous network solution based on years of innovative practices in the telecommunications field.
Overview of IP Network Intelligence System
Fig 1. shows the IP network intelligence system developed by ZTE. The bottom layer is the IP network layer, which needs the NE device to deliver the endogenous intelligence, such as self-awareness, self-configuration, self-recovery, and self-optimization.
The middle layer is the management and control system, and it is necessary to build an intra-domain intelligent closed-loop system. ZENIC ONE is ZTE's intelligent management and control system based on the cloud native platform. It contains awareness, control and intent engines, and combines the latest AI and big data technologies to generate a closed-loop system for sustainable learning.
The ZENIC ONE provides an intelligent and open API for the upper-layer system to support cross-domain global coordination and network application-based ecological closed loop.
Endogenous Intelligence of IP Network
Endogenous network intelligence allows real-time monitoring and analysis of IP NE data to best match software and hardware resources and conduct precise O&M for intelligent control. It delivers precise performance, status, and quality data for the upper-layer system, and allows flexible scheduling.
Awareness
The IP network needs to support the real-time awareness. The NE monitors its own software/hardware and the operational status, workload, and alarms of the services in real time, and offers data support for local or network-level closed-loop decision making. ZTE improves the NE awareness from the following aspects:
—Upgrade the out-of-band detection to the in-band one, and make IP traffic visible and measurable. IP IOAM can test the service quality of the IP network more accurately, and cut the number of detection packets in the network to improve the accuracy and efficiency of network awareness.
—Increase the granularity of network awareness with the data collection intervals improved from 30 seconds to 10 seconds, and then 1 second.
—Intelligentize network awareness. In the precise awareness scenario, AI is introduced to enable self-learning and self-adaption of granularity for awareness. This technology can minimize the amount of data collected by the network, and ensure the awareness of tiny network status changes while improving the overall efficiency.
Self-Configuration
ZTE improves the automatic deployment of the IP network from the following aspects:
—Open NE capability by supporting such interface protocols as NETCONF, PCEP, BGP SR Policy and BGP-FS.
—Support NE configuration rollback to improve the fault tolerance of automatic NE deployment. If the configuration fails to be delivered, it will automatically roll back. The running configuration can be rolled back to a set rollback point.
—Allow NE to support VPN and tunnel combined configurations and automated service deployment and optimization.
Self-Recovery
ZTE has made the following improvements in the intelligent diagnosis and recovery of faults:
—Make redundancy protection for key components, and support millisecond-level hardware switchover.
—Support fast detection of BFD link faults and provide a 50 ms detection mechanism for active/standby service switching so that the NE can switch services quickly and recover services automatically.
—Enable TI-LFA FRR to achieve fast traffic recovery.
—Support configuration of service escape paths. When the end-to-end service path fails, services can be quickly recovered by escaping to the protection path.
Self-Optimization
IP network self-optimization evolves from manual to intelligent and proactive optimization. ZTE has made the following improvements:
—Adjust device power.
—Support automatic fan speed adjustments. The temperature of NE boards and racks is monitored for self-judgment and decision-making.
—Dynamic sleep based on component load. When the component load is low, the system automatically determines whether to sleep the components to save energy.
Intelligent Management and Control System
In the current IP network O&M, service O&M and network troubleshooting are the most common scenarios. Service O&M automation and network troubleshooting intelligence can be enhanced to achieve autonomy within the IP network management and control domain.
Automatic Service O&M
Automatic Service O&M can be divided into two phases: deployment of automatic service provisioning and service maintenance and optimization.
The major problem in IP network service provisioning is low intelligence, which affects the service resource allocation speed. Moreover, service parameters are complicated and may be incorrectly configured.
In the IP network service provisioning (Fig. 2), the ZTE ZENIC ONE system automatically generates a service template according to the user's service requirements. The service is configured based on the service template. After the simulation succeeds, the system automatically sends the configuration to the device for service configuration verification.
It can be seen that the user only needs to pay attention to customer service requirements rather than professional network knowledge. The number of parameters that need to be concerned is greatly reduced.
When it comes to service maintenance and optimization, the ZENIC ONE system offers service self-healing and self-maintenance capabilities. The service awareness module can collect the service data in the network, and analyze the service operation information. If not as intended, the system optimizes services and modifies service configuration.
As network scale and traffic rise, service operation and protection routes will be unable to reach their optimal status. It is then necessary to optimize service routes and parameters, adjust service intents, design and simulate services again, and deliver optimized configurations.
On the other hand, as network services change, unbalanced network traffic has an increasing impact on the quality of the entire network, which needs to be detected and automatically optimized. The ZENIC ONE system can automatically monitor network traffic, analyze historical traffic and adjust network service bandwidth.
Intelligent Network Troubleshooting
There is great difficulty in identifying potential faults. Massive alarms overwhelm the monitoring system, and the network structure and service distribution are complicated. These make it hard to locate the root cause and solve the fault, and affect the O&M efficiency. To solve these problems, the ZENIC ONE has improved the intelligence in network monitoring and troubleshooting.
The management and control system divides network troubleshooting into three phases: fault identification, fault delimiting and locating, and fault handling and verification.
Fault identification relies on the intelligent sensing capability of the network to collect network status data and fault data from devices and the built-in intelligent analysis system to identify faults and potential troubles. It includes:
—Remote collection and big data analysis.
—Multidimensional fault analysis and tracing to quickly identify the root cause of a fault.
—Train data to create a fault model base.
Intelligent fault delimiting and locating is at the core of network troubleshooting. It supports precise scenario-based delimiting and locating, and can analyze and locate a specific fault within five minutes. If the detection module involves information exchanges associated with the device data, the fault can also be located within 15 minutes. The fault delimiting and locating support the following features:
—Hierarchical detection and fast fault delimiting.
—Automatic and visualized fault locating.
—Flexible policy as required by the users.
Fault handling and verification is the key of automatic closed-loop network troubleshooting, and allows automatic creation of systematic troubleshooting solutions. It includes:
—Automatically recommend multiple solutions and handling suggestions.
—Improve the level of fault repair automation.
The ZENIC ONE also supports simulation analysis suggested by network troubleshooting. Through simulating scenario-based network faults, it can find potential network troubles, and give a risk response plan before failure takes place.
Intelligent and Open IP Networks
The end-to-end intelligence of inter-domain services requires splitting the upper-layer system into different management and control systems. The open API of an intra-domain system is invoked to use the system intelligence. The upper-layer system performs end-to-end integrated analysis for intelligent O&M of cross-domain services.
The ZENIC ONE management and control system provides the upper layer system with intelligent and open capabilities of IP network and service data, supports northbound interfaces based on REST, SFTP, Socket, and SNMP protocols, and allows intelligent capabilities like service path provisioning, VPN service delivery, alarm, performance and resource obtaining, and notifications.
Summary
ZTE has been committed to exploring the evolution of network intelligence and automation and solving various challenges faced by traditional networks through innovative solutions. By practicing the self-intelligent network concept and introducing the latest technical solutions, ZTE will continuously increase network O&M efficiency and resource utilization, and improve the intelligence and sustainability of IP networks.
.png){kind=icon}
.png){kind=icon}
.png){kind=icon}
.png){kind=icon}