How to use autonomous networks for conflict resolution through intent APIs

Commercial context

CSPs occupy a unique position in the tech sector, in being crucial to the digital transformation of almost any industry – but their ability to monetize this singular opportunity is hampered by barriers to understanding the requirements they are asked to support. Matching the increasing sophistication of what CSPs have to offer with a potential customer’s systems and needs is a complex undertaking, and one which calls urgently for automation.

Intent-driven autonomous decision-making is the key here – specifically, through intent APIs. Intent APIs help businesses better understand the goals behind a customer's request or query by analyzing the context and meaning of the interaction intelligently, enabling more accurate and relevant responses and reactions, leading to improved customer experiences and business outcomes at the same time. ‘Intent’ in this context represents the desired outcomes, while autonomous domains are the systems that exchange intents to achieve those objectives. Intent-driven networks recognize that requirements regularly come into conflict, especially when they originate from different users or stakeholders, and seek to harmonize demands into optimized solutions. Simply put, they strive to achieve the best possible outcomes while still balancing conflicting requirements.

The solution

The Intent-driven autonomous networks - Phase III Catalyst is creating a more comprehensive understanding of intent APIs to help unlock the potential business benefits of using intent in autonomous networks, by assisting CSPs in identifying and integrating intent APIs relevant to their business objectives. The project will also pioneer the use of intent-based automation by connecting the TMF921 Intent API and other intent APIs with non-RDF based models defined by different SDOs. The Catalyst’s goal is in essence to support conflict resolution by supporting CSPs in contemplating and analyzing multiple variables, in a holistic way across domains, through five stages:

1. Measurement: determining the state of the system, with continuous observations and measurements made for every requirement in the intent
2. Assurance: determining if the system meets the stated requirements
3. Proposal: establishing viable solutions
4. Evaluation: deciding how closely the proposed solutions meet the intent requirements
5. Actuation: executing the solution devised to maximize the system’s utility level

This offers CSPs the chance to adapt their networks so that various requirements can simply be stated and then automatically translated into actionable insights – allowing them to introduce entirely new behaviours and services into their networks easily, without the need for cumbersome manual processes. The network can then be taught to continuously optimize itself based on changing conditions, adapting in real time according to the prescriptions set within the intents specified to propagate at all levels.

Now in its third phase, the Catalyst is adding new modules to achieve practicable autonomous decision-making for conflict management. Machine learning and AI algorithms will be used to analyze historical data, network behavior patterns, and contextual information to resolve conflicts based on the resulting insights. Reinforcement learning techniques will also be used to help network elements, agents, and controllers learn from feedback and maximize rewards. Where multifaceted conflicts arise – where for instance multiple autonomous agents or controllers need to make independent decisions – distributed consensus mechanisms such as voting algorithms or consensus protocols can be employed to achieve collective decisions based on predefined rules. Finally, an intent negotiation and collaboration module will enable involved entities to exchange intent information and work together on beneficial solutions.

Wider value

This project introduces, for the first time in the industry, an approach leveraging the concept of utility to meet these goals. In the context of autonomous systems, utility refers to the value or usefulness of a particular system or component to a user, or to the overall network. It is often used to measure the effectiveness or efficiency of a system or component in achieving a particular goal or meeting a specific need. As such, the AN project refers to utility as an aspect of intent as it describes "knowledge about what makes an outcome or situation preferential" (IG1253).

For example, an autonomous system that is designed to optimize energy consumption in a building may have high utility if it can significantly reduce energy costs and improve energy efficiency. Similarly, an autonomous system that is designed to facilitate communication and collaboration among a group of people may have high utility if it can effectively facilitate communication and improve productivity. As Project Champion Massimo Banzi, Senior Std. Manager at Telecom Italia SpA explains, “CSPs can provide services that optimize resource allocation and energy consumption, generate savings and provide guaranteed quality of service in the process, while fully meeting all requirements.” And with this Catalyst now successfully complete, the proof is there.