Y.3172 is an
ITU-T
The ITU Telecommunication Standardization Sector (ITU-T) is one of the three sectors (divisions or units) of the International Telecommunication Union (ITU). It is responsible for coordinating standards for telecommunications and Information Co ...
Recommendation specifying an architecture for machine learning in future networks including 5G (IMT-2020).
The architecture describes a
machine learning
Machine learning (ML) is a field of inquiry devoted to understanding and building methods that 'learn', that is, methods that leverage data to improve performance on some set of tasks. It is seen as a part of artificial intelligence.
Machine ...
pipeline in the context of
telecommunication networks that involves the training of machine learning models, and also the deployment using methods such as
containers
A container is any receptacle or enclosure for holding a product used in storage, packaging, and transportation, including shipping.
Things kept inside of a container are protected on several sides by being inside of its structure. The ter ...
and
orchestration
Orchestration is the study or practice of writing music for an orchestra (or, more loosely, for any musical ensemble, such as a concert band) or of adapting music composed for another medium for an orchestra. Also called "instrumentation", orch ...
.
A set of architectural requirements and specific architectural components needed to satisfy these requirements are presented. This includes i.a., machine learning pipeline as well as machine learning management and orchestration functionalities. Additionally, the standard describes the integration of such components into future networks including IMT-2020 as well as guidelines for applying this architectural framework in a variety of technology-specific underlying networks.
The Recommendation
Y.3173 builds upon Y.3172 by specifying a framework for evaluation intelligence levels of future networks.
Machine Learning pipeline
* 'SRC' (source): This node is the source of data that can be used as input to the ML pipeline.
* 'C' (collector): This node is responsible for collecting data from one or more SRC nodes.
* 'PP' (preprocessor): This node is responsible for cleaning data, aggregating data or performing any other preprocessing needed for the data to be in a suitable form so that the ML model can consume it.
* 'M' (model): This is a machine learning model, in a form which is usable in a machine learning pipeline.
* 'P' (policy): This node enables the application of policies to the output of the model node.
* 'D' (distributor): This node is responsible for identifying the SINK(s) and distributing the output of the M node to the corresponding SINK nodes.
* 'SINK': This node is the target of the ML output on which it takes action.
History
At the 2017
AI for Good summit the idea for a Focus Group on Machine Learning for 5G (FG-ML5G) arose, similar to the way the summit gave rise the
ITU-WHO Focus Group on Artificial Intelligence for Health
The ITU-WHO Focus Group on Artificial Intelligence for Health (AI for Health) is an inter-agency collaboration between the World Health Organization and the ITU, which created a benchmarking framework to assess the accuracy of AI in health.
Thi ...
. The Focus Group finalized the standard in May 2019 and submitted it to
ITU-T Study Group 13
The ITU-T Study Group 13 (SG13) is a statutory group of the ITU Telecommunication Standardization Sector (ITU-T) concerned with networks, infrastructure and cloud computing, including the networking aspects of mobile telecommunications. Example ...
(its parent body) for adoption as an ITU-T Recommendation, which happened in June 2019.
Based on Y.3172 ITU launched an AI/ML 5G Challenge, on implementing this architecture.
References
{{Reflist
External links
ITU-T Recommendation Y.3172ITU-T Focus Group on Machine Learning in 5GITU News MagazineAI/ML 5G Challenge
ITU-T Y Series Recommendations