Adopting intent: lessons from outside telecoms

The telecoms industry has a long history of going its own way, but increasingly it is embracing technology and expertise from IT and cloud computing to meet the cost, scale and agility requirements required by the digital economy.

Telecoms can learn a lot from the best practices other industries have adopted for how autonomous systems make decisions, and the relevance to intent-based telecoms systems of technologies used elsewhere to build decision-making systems. IG1253 is the proposal by the TM Forum Autonomous Networks Project (ANP) on how to implement intent in autonomous networks.

Jörg Niemöller, Expert of Analytics and Customer Experience at Ericsson and IG1253 guide lead and author, says several critical elements of the team’s proposal are unique within telecoms. But he adds: “Intent as a mechanism to carry requirements, build control loops and operate systems is not limited to telecoms.” Learning from use cases outside telecoms can help accelerate applications of autonomous systems because the needs will be very similar.

Adopting established technologies

TM Forum’s report Intent in autonomous networks explains why the concept of intent is essential to autonomous systems evolution. It also outlines the key concepts and building blocks of intent-driven automation that draw from expertise outside telecoms.

Niemöller says: “There are established technologies, such as ontology-based knowledge graphs and machine reasoning, that are not yet adopted by telecommunication standards and systems, although they are mature, well understood and used elsewhere. They constitute an entire branch of artificial intelligence, which was developed for machines to make sense of the information given to them and to preserve abstract knowledge in a way that it can be used by machines.”

Ignoring relevant technologies such as these could lead to disruption from other industry sectors; companies such as Google already use these technologies very effectively, for example.

Adoption of such useful technology underpins the ANP team’s recommendations for the development of: the intent interface (the API); intent common model (the basic vocabulary and semantics required to support intent management); and intent extension models (intent management domain and use case specific vocabulary and semantics) fundamental to intent-driven autonomous networks.

The team proposes using purpose-built tools based on RDF (Resource Description Framework) to implement abstract and dynamic concepts such as intent and support machine reasoning and logic inference.

Evolutionary, not revolutionary adoption

Technologies such as RDF will augment those such as Unified Modeling Language (UML), now used by telecoms developers for the open APIs used to enable interoperability and automation.

Niemöller notes: “While we certainly add new techniques in the definition of the intent API, we do not completely turn things upside down. It is more an evolution that keeps many established principles in place.”

If machines are ever to acquire true understanding and automate human-like decision-making, says Niemöller, “we must look beyond the established policy-driven implementations” because all expected situations cannot be anticipated and accounted for in advance. The good news is that because autonomous systems are not new outside of telecoms there is substantial expertise to tap as our industry adopts these essential approaches.

Download the report Intent in autonomous networks to find out more.