Artificial Intelligence and Machine Learning
By Gino Masini, Yin Gao, Sasha Sirotkin (Article first published in HIGHLIGHTS Issue 02, March 2021).
Title: Artificial Intelligence and Machine Learning in NG-RAN: New Study in RAN3.
5G brings more stringent requirements for Key Performance Indicators (KPIs) like latency, reliability, user experience, and others; jointly optimizing those KPIs is becoming more challenging due to the increased complexity of foreseen deployments. Operators and vendors are now turning their attention to Artificial Intelligence and Machine Learning (AI/ML) to address this challenge. For this reason, following RAN plenary approval, 3GPP RAN3 has recently started a new Release-17 study on the applications of AI/ML to RAN.
AI can be broadly defined as getting computers to perform tasks regarded as uniquely human. ML is one category of AI techniques: a large and somewhat loosely defined area of computer algorithms able to automatically improve their performance without explicit programming. AI algorithms were first conceived circa 1950, but only in recent years ML has become very popular partly due to massive advancements in computational power and to the possibility to store vast amounts of data. ML techniques have made tremendous progress in fields such as computer vision, natural language processing, and others.
ML algorithms can be divided into the following types:
- Supervised learning: given a training labeled data and desired output, the algorithms produce a function which can be used to predict the output. In other words, supervised learning algorithms infer a generalized rule that maps inputs to outputs. Most Deep Learning approaches are also based on supervised learning.
- Unsupervised learning: given some training data without pre-existing labels, the algorithms can search for patterns to uncover useful information.
- Reinforcement learning (RL): unlike the other types, which include a training phase (typically performed offline) and an inference phase (typically performed in “real time”), this approach is based on “real-time” interaction between an agent and the environment. The agent performs a certain action changing the state of the system, which leads to a “reward” or a “penalty”.
Perhaps the most obvious candidate for AI/ML in RAN is Self-Organizing Networks (SON) functionality, currently part of LTE and NR specifications (it was initially introduced in Rel-8 for LTE). With SON, the network self-adjusts and fine-tunes a range of parameters according to the different radio and traffic conditions, alleviating the burden of manual optimization for the operator. While the algorithms behind SON functions are not standardized in 3GPP, SON implementations are typically rule-based. One of the main differences between SON and an AI-based approach is the switch from a reactive paradigm to a proactive one.
Stay tuned for further updates as the study progresses in RAN3, or consider joining us in our journey into the “uncharted” territory of AI/ML in NG-RAN.
- Study item can be folllowed by looking at the SID's Acronym: FS_NR_ENDC_data_collect
- TR 37.817 - Study on enhancement for Data Collection for NR and EN-DC (Early version only at June 15, 2021).
Contact for this article: