“With the accelerated development of the digital transformation of the whole society, 5G has become an indispensable next-generation network infrastructure. 5G poses higher challenges to the operation and maintenance of the bearer network: the hybrid networking of 4G/5G networks, the wide application of high-speed mobile broadband services, and the large-scale mobile Internet of Things make the scale and service capacity of the bearer network explode. The structure presents multi-dimensional complexity, and the complexity of operation and maintenance is beyond human reach; the industry application requirements of the business are diversified, and users have higher requirements for the quality and efficiency of network services; 5G construction investment is huge, and the overall revenue growth of the communication industry is slow. Network operators face lower Capex and Op

“

With the accelerated development of the digital transformation of the whole society, 5G has become an indispensable next-generation network infrastructure. 5G poses higher challenges to the operation and maintenance of the bearer network: the hybrid networking of 4G/5G networks, the wide application of high-speed mobile broadband services, and the large-scale mobile Internet of Things make the scale and service capacity of the bearer network explode. The structure presents multi-dimensional complexity, and the complexity of operation and maintenance is beyond human reach; the industry application requirements of the business are diversified, and users have higher requirements for the quality and efficiency of network services; 5G construction investment is huge, and the overall revenue growth of the communication industry is slow. Network operators are under enormous pressure to reduce Capex and Opex. Faced with the challenge, ZTE believes that realizing network automation is an inevitable trend of network evolution, and thus launched the bearer network management and control fusion automation system.

Bearer management and control fusion automation system

The automation technology reconstructs the entire network architecture and operation system to improve productivity, breaks the closed structure of the traditional network, and divides the network into the operation layer, the management layer, and the network element layer. As the core product of the management and control layer, ZENIC ONE, the integrated bearer management and control system, is the backbone linking the previous and the next, and the brain of network automation.

ZENIC ONE consists of an intent-driven engine, an automatic control engine, and a network-aware engine. The three engines respectively realize the construction and maintenance of intent, orchestration and control of services, and real-time network monitoring/analysis. ZENIC ONE provides a full life cycle management and control model from planning, deployment, assurance, to optimization, and forms an automated closed loop in each link. Intelligence endows network management and control with new capabilities, further improving the automation level of management and control systems. Telemetry provides millisecond-level data collection and provides real-time network status perception capabilities for management and control systems; big data technology realizes data preprocessing and provides high-quality data foundation for AI analysis and reasoning; AI technology is introduced to generate various data through training data. The model required for the scenario, through model inference, can finally realize the accurate identification and prediction of potential risks such as network faults and abnormal network traffic. The AI+BigData platform runs through the entire system, provides support for the automation of the management and control system, improves network quality as a whole, simplifies the operation and maintenance process, and reduces the complexity of operation and maintenance.

Figure 1 Full life cycle intelligent operation and maintenance

ZENIC ONE is developed based on micro-service architecture. It uses cluster + virtualization technology to achieve centralized control and supports flexible expansion of management capabilities. It can manage and control a super-large network with a maximum of 300,000 equivalent network elements, and its network management and control scale is at the forefront of the industry. Through the centralized control of the ultra-large network, the end-to-end unified scheduling and optimization of the entire network services can be realized, and the global resource utilization rate can be optimized.

As one of the few manufacturers in the industry that provides full products, ZTE ZENIC ONE realizes a set of management and control system to cover all scenarios of 4G/5G bearer networks such as IP/IPRAN, OTN, PTN/SPN, IP+optical, network slicing, cloud-network integration, etc. Significantly saves resource overhead, reduces OPEX, and improves user experience.

Typical functions of bearer control automation

Automation is a systematic project involving all aspects of the network life cycle. This article only extracts several typical functions of automatic network construction, centralized SDN control, network slicing, intelligent configuration inspection, and traffic prediction to illustrate the automation capabilities of ZENIC ONE.

automatic network construction

In the traditional operation and maintenance mode, network construction/expansion requires manual careful network design and planning, and devices are activated point by point on the site. The ZENIC ONE R&D network is automatically constructed to solve the above pain points. First, it supports the automatic online of DCN. After the equipment hardware is installed, it no longer needs to manually enter the site to open the site. It can automatically identify neighbor parameters, and automatically discover network elements, boards, and links, and automatically generate topology. . Second, the management and control system can make global planning for logical resources such as routing domain, IP, and network bandwidth, and form different configuration templates for different scenarios. Users can automatically generate configuration data based on existing templates and issue devices with one click to quickly build basic network data.

The automatic construction of the ZENIC ONE network improves the network opening efficiency, reduces the complexity and operation steps of opening, and reduces the probability of human error.

SDN centralized control

An important component of automation is centralized SDN control, which is divided into path control and tuning. ZENIC ONE senses network topology changes through BGP-LS, realizes network control through PCEP/BGP/Netconf, and supports SDN capabilities such as IP/MPLS, SR-TP, SR-TE path calculation, and path restoration. ZENIC ONE adopts an optimization algorithm to support various scenarios such as link congestion re-optimization, global traffic balancing, delay optimization, and topology change re-optimization. The management and control system can also analyze the collected data, generate an optimization strategy according to the optimization goal, and issue devices for network optimization.

The core capability of SDN control is algorithm, for which ZTE has developed an intelligent iTE algorithm engine. As the core Module of SDN centralized control, iTE integrates many self-developed technologies, such as path packing, parallel resource deduction, and automatic stack depth optimization. capacity and network resource utilization.

network slice

Network slicing is one of the most distinctive features and advantages of 5G, which enables the physical network to be logically divided into multiple different types of virtual networks to meet the needs of different application scenarios. The bearer management and control system supports slice management of the bearer network and control of the slice network, and provides an open interface to realize the automatic creation of slices.

A typical slice creation process is shown in the following figure:

Figure 2 Slice creation process

When the bearer management and control system receives user-oriented SLA-level parameters and slice application scenarios, the built-in TN NSSMF module first obtains the planning data of the slice to be activated according to the scenario, and at the same time automatically obtains resource information from the resource pool, and requests service management for service configuration; secondly , the business management module dynamically allocates resource information according to business configuration requests and parameter requirements, obtains data from the outside through static presets or arrangement based on the TOSCA model, completes the configuration information, and sends devices to build slices. After a slice is created, the management and control system can monitor the slice in real time and ensure the service quality of the entire slice in real time.

Based on the above process, ZENIC ONE, together with our company’s core network and wireless, completed multiple slice PoC verifications in large Ts such as China Mobile and Orange, laying the foundation for the commercial deployment of slices.

Smart Configuration Check

Service assurance includes many aspects, and checking the correctness of network configuration is an important part. The workload of traditional configuration inspection is huge and error-prone. Implementing automatic inspection of network configuration can greatly reduce operating costs.

ZENIC ONE extracts the configuration feature structure of the device from the existing network configuration data, forms the role fingerprint of the device, and compares the new network configuration with the corresponding fingerprint to identify the configuration risk, thereby realizing the configuration intelligent check function, which is helpful for network opening and Follow-up maintenance provides a guarantee.

The role fingerprint refers to the important parameter characteristics of a specific type of service. For example, the L3VPN service has parameters such as port, IP address, and tunnel. The correct value range/value of each parameter of the L3VPN service in different scenarios constitutes the L3VPN service in this scenario. under the character fingerprint. The configuration parameters of the device depend on the type of the device, the location of the device in the network (access, aggregation, core), service bearer characteristics, network topology, and connectivity characteristics. Therefore, each type of device has multiple role fingerprints. The principle of configuring the smart check function is shown in the following figure:

Figure 3 Device Configuration Role Fingerprint

First, extract massive configuration data from existing network devices, use artificial intelligence algorithms (such as NLP), build a device hierarchical model (network-device-service-configuration commands-configuration parameters), and form a knowledge map of device configuration based on association rules at different levels , and on this basis, each role fingerprint is formed through cluster analysis. Secondly, automatically check each network device, and find out possible configuration abnormalities by matching the role fingerprints corresponding to the device. Finally, the extracted abnormal configuration is checked, and if it is finally determined that the configuration is normal, it can be stored and used for fingerprint update.

The configuration intelligent inspection function can reduce the traditional configuration inspection workload of 90 man-days to 7 man-days, and at the same time increase the fault identification rate to 85%, effectively reducing manpower investment and reducing potential network configuration risks.

Traffic forecast

By predicting the future network traffic, the load of the future network is predicted. Traffic prediction provides support for network capacity planning, network congestion risk warning, and fault detection/simulation, and is an indispensable basic capability in network intelligence.

The network traffic has the characteristics of suddenness, holiday effect, tidal effect, etc., and the network traffic model is different on the granularity of different network resources (ports, links, ring networks), which is based on the characteristics of network traffic. ZENIC ONE uses a variety of technologies to improve the accuracy of prediction. First, it uses Telemetry and Inband-OAM technology to achieve accurate collection of sample data; secondly, it relies on AI platform to clean, convert and preprocess data, and integrate time, business, and network resource types. Feature extraction and model selection are carried out in multiple dimensions, so as to establish accurate prediction models for different network resources; finally, an appropriate evaluation model is established for each prediction model to realize the closed-loop of the whole process of traffic prediction, and realize the continuous optimization of the prediction model through machine learning technology. Improve forecast accuracy.

Through the comprehensive application of various technologies, ZENIC ONE intelligent traffic forecasting will play an important role in improving network service quality and tapping potential business opportunities.

The development of automation of management and control systems will gradually free people from complicated network operation and maintenance, greatly reducing operating costs and maximizing the release of network potential. ZENIC ONE developed by ZTE has been successfully commercialized/trial-commercialized in many large T operators at home and abroad, and has cooperated with many operators and research institutions to develop dozens of innovative application functions related to automation; and was upgraded in the GlobalData rating in April 2019 As “Very Strong”, it won the Product Innovation Purple Gold Award at the Artificial Intelligence Summit in July, and the “Best Network Intelligence” award at the World Broadband Forum in October. ZTE will continue to work closely with partners to promote the continuous improvement of network automation capabilities, and ultimately achieve network autonomy and network autonomy.