What kind of network slicing does 5G require?

Everyone should be familiar with network slicing.

That is, based on the SLA requirements of different industry applications for network bandwidth, latency, security, reliability, geographical coverage, etc., multiple isolated, secure, and SLA-guaranteed logical networks are "cut" out on demand from the end-to-end network infrastructure.

Network slicing has changed the past "one-size-fits-all" and "best-effort" network model. It provides differentiated services to various industries through dedicated, isolated and SLA-guaranteed "slices", thereby promoting massive vertical industry applications such as Internet of Vehicles, smart factories, and drones, and enabling the digital transformation of the entire society.

This is somewhat similar to urban traffic roads. In the past, there was no solid line separation, and various vehicles competed for the road, causing traffic congestion. Now, according to different vehicles and operating modes, diversion and isolation are carried out, such as setting up BRT rapid bus lanes and non-motorized vehicle dedicated lanes, which greatly improves traffic efficiency.

With network slicing, the network will be transformed from a "pipeline" to a "network infrastructure that integrates industry services". The operator's business model will also be expanded from "selling traffic" to consumers to "selling network capabilities" and "selling services" to various industries, thereby maximizing the value of the network and increasing revenue sources.

No slicing, no 2B. There is no doubt that network slicing is one of the most innovative technologies of 5G.

But faced with a huge and complex network infrastructure, how should operators “slice” it? How should they “cut” it?

Is it like cutting bread? Cut one slice per person according to their appetite?

Although this is also very vivid, the piece of bread is the network, computing and storage resources, and the slices cut are the resources required for specific communication services, but in fact, the process behind network slicing is not that simple...

(Network Slicing Service Process)

As shown in the figure above, the network slicing service process is roughly divided into the following steps:

1) Vertical industries raise demands for communication services

For example, a power company would lease a “slice” from an operator to provide network services for its drone inspection business.

2) The slice management platform “translates” communication service requirements into specific SLA requirements for network slices

SLA refers to the specific bandwidth, latency, and reliability required to support drone inspections, how many drones are connected, and the inspection area.

3) Slice design

After determining the SLA requirements, the network functional components, network resources, connection topology, etc. required for network slicing are designed according to the requirements. Since network slicing design is very complex, basic templates for various network slices are usually defined first, and then modified and optimized based on the templates.

4) Slice instantiation

It means putting slices online and opening them, which means deploying the resources, network functions, connection topologies, etc. required for network slicing on the virtual network infrastructure through orchestration management, and activating them online to provide slicing services to vertical industries.

5) Slice Operation and Maintenance

This means monitoring and adjusting network slices during the operation phase of network slices to ensure slice SLA in real time. Slice operation and maintenance can be open to vertical industry tenants so that they can understand and optimize slice networks in real time, as well as perform secondary development of operation and maintenance functions.

It is not difficult to see from the above process that network slicing is not like "cutting bread" but more like "customizing bread". After receiving the customized requirements of customers, the baker must first understand the customer's needs, and then mix the basic materials such as flour, water, yeast, butter, eggs, salt, sugar, etc. in appropriate proportions according to the requirements, and then ferment, shape and bake.

This process tests your skills and experience. The proportions of raw materials must be precise, and you can’t just take your time. The customer is already hungry, so you have to finish the product quickly and deliver it to him.

In this case, what kind of network slicing solution does 5G need?

As a major contributor to the 3GPP standard network slicing architecture, Huawei began research on slicing in 2015, and carried out application pilots in multiple industries in cooperation with global mainstream operators such as Deutsche Telekom and China Mobile, and launched the Slicing Mall in 2018. Based on its deep accumulation in 5G slicing technology and applications, Huawei provides the industry's only fully automatic multi-dimensional intelligent dynamic slicing solution. Let's take a look at the features of this solution.

Cut finer

The raw materials for making bread are flour, water, yeast, butter, etc. What are the "raw materials" for network slicing?

Take the 5G core network as an example. As we all know, NFV decouples virtualized network functions (VNFs) from traditional dedicated equipment, and the 5G core network is designed for cloud native, drawing on the "microservice" design concept and adopting the SBA architecture, which further decomposes VNFs into smaller-granular, stateless, relatively independent and flexibly callable network function service modules, so that network functions can be modularly "assembled" and network slices can be flexibly customized according to business needs.

At the same time, the control plane (CP) and user plane (UP) of the 5G core network are completely separated, supporting distributed UPF and edge computing deployment that can be flexibly sunk, making it closer to the data source, thereby providing low-latency and high-reliability network services for industry applications such as smart manufacturing and unmanned driving through end-to-end network slicing.

These are the “raw materials” for network slicing.

Network slicing is based on network cloudification and functional modularization, and flexibly designs and creates network functional components, network resources, connection topologies, etc. according to specific business needs.

In bread making, from raw material selection and mixing to fermentation, shaping, and baking, each step must be detailed enough to avoid wasting materials and satisfy customers. The design and deployment process of network slicing is similar, except that the composition and business needs of mobile networks are much more complex than bread, which requires more detailed work. For example, industry needs include multiple dimensions such as network performance and service functions. Whether a network slice can meet the needs of multiple dimensions depends on whether it can achieve the output and configuration of effective resources according to the requirements of multiple dimensions.

Huawei's innovative multi-dimensional intelligent algorithm breaks through the limitations of traditional limited-dimensional resource calculation. It can accurately calculate SLA based on global real-time data such as network load and topology, and achieve dynamic SLA verification and cost optimization through online model optimization and other algorithms. It outputs network resource models and realizes the conversion of multi-dimensional network requirements into accurate models.

At the same time, Huawei effectively decomposes network functions into microservices, so that network resources can be flexibly orchestrated based on smaller-grained microservices, which is more efficient and more real-time.

Cut faster

It is expected that operators will design and maintain hundreds or even thousands of network slices in the future. If they rely on traditional manual methods to investigate and design the slice requirements of different industries, as well as deploy, test and accept each network element, it will take several months to launch a new service. Obviously, this cannot meet the differentiated needs of various industries.

Therefore, in order to quickly respond to market business changes, automation is a key component of network slicing.

In this regard, Huawei's fully automatic multi-dimensional intelligent dynamic slicing solution is based on a fully automatic programmable pipeline engine, which can realize online research and network design, as well as one-click automated deployment. Through Huawei MAE-CN (MBB Automation Engine, Mobile Network Automatic Operation Engine), online automatic orchestration and deployment are built to achieve weekly slicing activation.

According to different business scenarios and security isolation requirements, Huawei's fully automatic multi-dimensional intelligent dynamic slicing can also build a variety of slicing modes, which can be flexibly combined to meet the demands of differentiated scenarios and business models. The private network isolation mode meets the needs of networks such as power grid differential protection that have high requirements for latency and jitter, deterministic SLA requirements, and high security isolation requirements. The resource isolation mode meets the needs of scenarios with lower isolation requirements, such as video surveillance and live broadcast. This requires the control plane to implement slice-level user management and the forwarding plane to implement slice-level bandwidth management, and decide whether to share some or all resources according to the degree of isolation requirements.

Cut more accurately

From the perspective of network slicing architecture, multiple end-to-end network slices can be cut horizontally, and each end-to-end network slice is divided into wireless access network sub-slices, bearer network sub-slices and core network sub-slices vertically. A terminal may access multiple slices at the same time, which is very complex.

During the operation of slices that meet the multi-dimensional needs of the industry, the KPI indicators may change due to changes in complex network conditions, thus deviating from the SLA requirements of the slice itself, causing deviations in user experience and even industry application failures. To achieve real-time end-to-end SLA guarantees, it is necessary to implement segmented monitoring of multiple network indicators and be able to provide preventive guarantees.

Huawei's fully automatic multi-dimensional intelligent dynamic slicing uses a lightweight monitoring solution to monitor the number of slice users, throughput, user-level bandwidth, latency and other indicators in real time, and segment latency end-to-end to achieve slice-level and user-level multi-dimensional monitoring, and SLA status accurate perception in seconds. And through the collection of end-to-end multi-dimensional data, intelligent big data analysis and full-network collaborative resource scheduling and optimization are carried out to achieve an AI-based SLA dynamic optimization closed loop to ensure the differentiated and deterministic network requirements of various industries.

More convenient to use

All industries generally have a need for self-service in the use of slices. For example, manufacturing plants need to monitor slice status and key KPI indicators to ensure that robots on the production line receive instructions online and can work under the specified bandwidth and latency requirements to ensure smooth production. New terminals connected to the slice network can be activated and signed independently. If multiple slices are purchased, the slice associated with the terminal needs to be modified according to business requirements.

Huawei's fully automatic multi-dimensional intelligent dynamic slicing provides a 2B one-stop operation platform, providing key industry self-service capabilities including self-operation and self-maintenance. Enterprises use a unified operation platform to manage slices in a decentralized manner. It provides one-stop self-operation from product ordering to service activation, enabling product customization, self-ordering, and group self-management. At the same time, it provides KPI monitoring of the number of users, bandwidth, latency, and other key operation and maintenance services for purchased slices, as well as 5G LAN management, diversion strategy configuration, APP management, network troubleshooting, and other key operation and maintenance services.

In short, from the perspective of Huawei's fully automatic multi-dimensional intelligent dynamic slicing solution, an excellent network slicing solution must have the following capabilities:

• Cut finer

Accurate modeling can be achieved based on multi-dimensional intelligent modeling capabilities and on-demand orchestration at the microservice granularity.

• Cut faster

Based on a fully automatic programmable pipeline engine, online research and network design, as well as one-click automated deployment, are realized.

• More accurate cutting

Based on the lightweight solution, slice-level and user-level KPI indicators are monitored in real time, segmented and delimited, and network resources are adjusted in real time to ensure slice SLA requirements.

• More convenient to use

Based on a unified portal, it provides industry users with slice-level decentralized management and domain-based management, making online operations and maintenance open and visible.