Generations of Mobile Standards

Intent Driven Management

Jul 27,2023

Enabling intelligence and autonomation for 5G Advanced Networks

By WG SA5 - Authors Ruiyue Xu (Huawei), Mark Scott (Ericsson), Stephen Mwanje (Nokia)

5G advanced networks bring more operational complexities due to the large number of devices and diversity of services. Network automation/intelligence is one of the important topics for 5G advanced networks, which aims to reduce the operating expenditure (OPEX) and improve the service experience to enable various vertical industries (e.g. autonomous vehicle, smart city) through a variety of intelligence and automation mechanisms. As technologies are evolving and the level of complexity explodes, the need for abstraction of the operational requirements, as supported by intents, becomes more apparent.

An intent is an expression of the desired state of a system used to describe an intended network or service. Intents do not define specific network or service configuration, nor do they prescribe management tasks to be performed by a system. Intents allow customers to request networks and services without detailed knowledge of how they will be provided. Inherently, this assumes that the system can learn the behavior of networks and services and use intelligence and automation mechanisms to fulfil the requests expressed via intents. This not only relieves the consumer of the burden of knowing implementation details but also provides flexibility allowing the producer to explore alternative options to find optimal solutions.

Intent concept

An intent is a set of expectations including requirements, goals, conditions and constraints given to a 3GPP system, without specifying how to achieve them. The main characteristics of an intent are:

  • An intent is typically understandable by humans, and needs to be interpreted by a machine without any ambiguity.
  • An intent focuses on describing "what" needs to be achieved, and not "how" the outcomes should be achieved. This is in contrast to a rule (focus on "how") which specifies the explicit logics or formula to be executed, and to a policy (focus on "what" + "how") which specifies the action(s) to be taken and the conditions under which they should be taken.

intent01

Figure 1

  • Any expectations expressed by an intent are agnostic to the underlying system implementation, technology and infrastructure. For example, an area can be used as object in the expectations expressed by an intent to achieve system implementation, technology and infrastructure agnosticism.
  • An intent needs to be quantifiable such that the fulfilment result can be measured and evaluated via network data including measurements and KPIs.

To support different roles related to 5G networks and network slicing management defined in 3GPP, different intents could be considered, being used for supporting different interactions, specifically:

  • Intent-CSC: from Communication Service Customer (CSC) to the Communication Service Provider (CSP) to express properties of a communication service, e.g. ‘Enable a V2X communication service for a group of vehicles in certain time with low latency’.
  • Intent-CSP: from CSP to a Network Operator (NOP) to express properties of the CSP’s desired network, e.g., ‘a network slice supporting V2X communications.
  • Intent-NOP: from NOP to a Network Equipment Provider (NEP) to express characteristics of a RAN and/or 5GC network, e.g., specifying ‘coverage requirements and UE throughput requirement in certain area'.

intent02

Figure 2

Intent also can be used for management and control of closed-loop automation, which means the intent can be translated to policies and management tasks that the MnS producer needs to execute for the closed-loop automation. In the intent driven management approach, the mechanisms that the MnS producer uses to achieve closed-loop automation to satisfy the intent is the implementation of the MnS producer and shall not be standardized.intent03a

Figure 3

Intent driven scenarios

Following are the typical intent driven scenarios defined in TS 28.312 and investigated in TR 28.912.

Table 1

Scenarios defined in TS 28.312

Scenarios investigated in TR 28.912 and to be defined in TS 28.312

  • Intent expectation for delivering radio network
  • Intent expectation for delivering radio service
  • Intent expectation on coverage performance to be assured
  • Intent driven approach for RAN energy saving
  • Intent containing an expectation on RAN UE throughput performance to be assure
  • Intent driven approach for radio capacity assurance
  • Intent containing an expectation for delivering a service
  • Intent driven approach for 5GC management, specifically 5GC network delivering

 

Intent driven MnS

Intent support in 3GPP is provided via an Intent Driven Management Service (MnS). As with other 3GPP defined MnSs in Service Based Management Architecture (SBMA), an intent driven MnS Producer (provided by an intent driven system) allows its MnS consumer to express intents for managing the network and services. Intent allows an MnS Producer to fulfil the intent and provide the result/information about the intent fulfilment information. Internally, the MnS Producer’s intent fulfillment includes translating the received intent to executable actions such as performing service or network management tasks or identifying, formulating, and activating service or network management policies.

The intent driven MnS solution specified by 3GPP (TS 28.312) is a simplified and flexible interface, which allows for extensions to support any new scenarios/services. The intent driven MnS solution uses the “model driven approach” in SBMA, which decouples the operations from the intent model. In the solution in TS 28.312 the following are defined:

  • A unified set of management operations and an intent information model (with both information model and data model defined) to support intent lifecycle management. Therein, an intent is a set of intent expectations, each containing a set of expectation targets with optional contexts possible for each expectation target, intent expectation or the intent itself.

Intent Lifecycle Management

Intent operations

Intent Model

  • Create an intent

createMOI operation

 intent tablefig2
  • Activate an intent

modifyMOIAttributes operation

  • De-activate an intent

modifyMOIAttributes operation

  • Delete an intent

deleteMOI operation

  • Modify an intent

modifyMOIAttributes operation

  • Query an intent

getMOIAttributes operation

 

  • Intent Expectation instances (Including RadioNetworkExpectation and ServiceSupportExpectation defined in TS 28.312) to support the typical intent driven scenarios. The following is one concrete example for intent instance contain RadioNetworkExpectation with YAML format to support RAN UE throughput assurance use case.
  intent04b
Figure 4

Based on the intent solution in TS 28.312, more intent driven management features have been investigated in the Rel-18 study on intents for mobile networks (TR 28.912). The new features can be grouped into 2 categories:

  • Improvements to intent handling capabilities, specifically support for an intent report, the management of Intent conflicts, feasibility checking for Intent fulfilment, discovery and obtaining of Intent handling capabilities.
  • Enhanced enablers for Intent Fulfilment, which include the testing of intent-driven MnS Capabilities, the mapping of intents to machine learning capabilities, intent-driven SON orchestration as well as Intent-driven for Management Data Analytics. 

Collaboration with other SDOs

The 3GPP defined intent management service is aligned with, and can be integrated with, similar mechanisms being defined across the industry to provide a comprehensive and interoperable solution. Specially, TR 28.912 investigated and concluded the solution for deployment scenarios for intent interfaces and provide guidelines for the transformation functionality between TM Forum intent management API for intent-CSC to 3GPP intent driven MnS for intent-CSP.

intent05aintent05b

Figure 5

Way Forward

Intent driven management is a key enabler in increasing the autonomy and efficiency of 5G advanced networks, not only for the 5G advanced networks themselves but for those who interact with them. As 5G advanced networks with automation/intelligence become increasingly self-sufficient, able to manage their own resources in increasingly efficient ways, the ways in which we interact with 5G advanced network must change. Network operators, and their customers, will increasingly shift their focus to ‘what’ they wish to achieve from their networks and services. 3GPP SA5 will continuously develop standards for intent driven management solution and welcomes any new requirements for intent driven management from the industry.

References

Intent Driven Management Service (IDMS) Study/Work Items in Rel-17 and Rel-18:

Release

UID

Work Item name

Specification

Acronym

Rel-17

810027

Intent driven management service for mobile network

TS 28.312

IDMS_MN

Rel-18

940046

Study on enhanced intent driven management services for mobile networks

TR 28.912

FS_eIDMS_MN

 

Specification References:

The common specifications for all 5G management services are also used for the intent driven management, including the SBMA and corresponding generic management services.

  • TS 28.533: " Management and orchestration; Architecture framework".
  • TS 28.532: " Management and orchestration; Generic management services ".

Specific intent driven management specifications in release 17 and 18 are the following:

  • TS 28.312: "Management and orchestration; Intent driven management services for mobile networks".
  • TR 28.912: "Study on enhanced intent driven management services for mobile networks".