Skip to main content

Harmonizing the use of intents across network and service management

A hand reaching out to touch a digital circle.

We have all experienced how mobile networks, especially the radio access network (RAN), have increased in complexity over the generations. I think, for instance, of the additional mechanisms needed to support multiple service types, higher frequency bands, and network slicing. Both operators and vendors are looking for ways to manage this complexity, and automation is clearly part of the answer. However, automation can create even further complexity (and costs), e.g., when an automation function for one KPI degrades another KPI. So, we need to smartly simplify how automations are generated both for operators and the enterprises that consume the service.

The solution is the intelligent use of intents, in what may be termed ‘intent-driven management’ or ‘intent-based management’, which enables a service consumer to request a desired state from a service producer (the network) by only stating the desired outcomes and not specifying how to achieve them. By abstracting and hiding the complexity, intents offer improved operability at lower cost, and they even enable non-experts to configure wide area networks.

The main question is where exactly can intents be used in the network and service management context? I believe that intents are usable throughout network and service management to manage and control network-related business processes, services and resources.

The concept of intent-based management

Consider the following scenario: Alice wishes to manage aspects of a tree on her property, ensuring, for instance, that the tree’s branches don't scratch her house. In the non-‘intent-based world’, Alice might use a chainsaw to cut off the branches herself or hire someone and tell them exactly what to do (i.e., cut the branches). In the intent-based paradigm, Bob exposes an arboriculture management service to Alice, so that Alice only states her desired outcomes to Bob (i.e., “I don’t want the house to get scratched by the tree”). Bob then figures out "how" to realize the desired outcome.

Figure 1. Alice tells Bob her desired outcomes as intent-expressed on the configuration parameters (A) or on the performance metrics (B).

Figure 1

Alice only states "what" is needed and not "how" to get it done, which are considered implementation details to be decided by Bob. Note that Alice may add extra constraints to be considered, e.g., that "I should retain my summer shade in the garden". The benefits are that there is no risk of Alice dictating a sub-optimal solution; she doesn’t need to know or explain technical details of the solution; and there’s no ambiguity since only the intent is shared. Similarly in network and service management, the consumer simply expresses the desired outcomes relating to configuration (Figure 1A) or performance features (Figure 1B) of the network, as well as their services or business goals. The service producer decides what to do, to realize those outcomes.

Applying intent-based management in telecom

Since intents simply state the desired state of a system or object, they could be applied to any object that has state. Accordingly, intents can be applied to any management layer in the network, as further illustrated by Figure 2.

Consider, the example in Figure 2, where a business customer needs the "cheapest internet connection for gaming for his client”. This business intent can be broken down into different intents at each of the subsequent management layers. For instance, the business management layer may not be able to define through a service order what exact service is needed to realize the "cheapest internet connection for gaming". Instead, expressing the requirement as a service intent can be more realistic, in this case as "a symmetric high-speed internet service at 20Mbps with roundtrip latency not exceeding 10ms".

Figure 2. Application of intents at all the different management layers

Figure 2

Standardization opportunities

Maximizing the value of intent-based management requires intent-driven interfaces to be standardized to ensure commonality in the use of intents across the entire network, including mobile, core, transport, fixed access and cloud resources. Standardization is needed in different standard development organizations (SDOs) as illustrated by Figure 3, i.e., to specify how intents shall be expressed to request the corresponding resources, services or products defined by that SDO. Some work has started in many of the SDOs, with a stable baseline available in the 3GPP network management group (SA5). The ORAN-ALLIANCE is making progress towards a first study report. Broadband Forum (BBF) and IEEE are only at the start of the discussion, e.g., with only a high-level description of intents provided by IEEE but no specifications yet. Similarly, the open-source communities for cloud resources and operations management (such as Nephio) are yet to open work on intent-related control.

Figure 3. Interfaces and standards organizations for standardizing intents

Figure 3

Collaboration is needed between the different SDOs to ensure intents can be used across the entire network. To ease these interface discussions, the SDOs should, at a minimum, apply the same meta model for the intent-driven management interfaces from which they derive their domain-specific models. Nokia has developed such a meta model and supported its standardization in 3GPP, wherein an intent is an information object stating the desired target outcomes on one or more entities or objects of the same type. 3GPP has accepted this meta model for 5G-Advanced, using it to derive the cellular-resource-specific models, which as illustrated by Figure 4, provide domain-specific values for the generic containers of the meta model.

Figure 4. Example of an intent as a structured object

Figure 4

It would be good if such a meta model is used as the baseline for other SDOs. Nokia is committed to supporting this. Moreover, it is also critical that solutions enabling intent-based management at all management layers are developed to guarantee the promised abstraction, flexibility and full scope of opportunities for business and service innovation. In line with this, Nokia has already developed pre-standardization solutions for managing fixed-access and IP/optical resources, while solutions for end-to-end service management are under development, as evidenced by the Intent Catalyst project that we are driving in the TM Forum (TMF). Nokia has the proven technical expertise and the ambition to drive the industry to successfully adopt intent-driven management solutions across the entire network. Intent-based management will enable communication service providers to create faster new value propositions and to minimize potential degradation of the end-to-end network.

Stephen Mwanje

About Stephen Mwanje

Stephen S Mwanje is a Research Project Manager leading "Network Automation" research and standardization efforts in Nokia Standards. He is an inventor or co-inventor on more than 70 patent applications, a co-author on more than 30 peer-reviewed publications and the main author of the book "Towards Cognitive Autonomous Networks: Network Management Automation for 5G and Beyond". He received his Ph.D. (Dr. Ing.) degree with summa cum laude at the Ilmenau University of Technology, in Germany. He is currently driving standardization of intent and AI/ML management interfaces in 3GPP SA5 while also contributing to harmonizing SA5 standards with TMF, ORAN and other related organizations.

Connect with Stephen on Linkedin

Article tags