.. This work is licensed under a Creative Commons Attribution 4.0 International License. .. SPDX-License-Identifier: CC-BY-4.0 .. Copyright (C) 2020 Nordix Requirements for the Non-RT RIC project ======================================= Find detailed description of what Non-RT RIC is on this `page`_. .. _page: https://wiki.o-ran-sc.org/display/RICNR/ There are functional requirements emitted by O-RAN WG2 for the Non-RT RIC, which are the following: #. Non-RT RIC shall support data retrieval and analysis; the data may include performance, configuration or other data related to the application (recommended data shown in required data section for different use cases). #. Non-RT RIC shall support relevant AI/ML model training based on the data in 1. for non-real-time optimization of configuration parameters in RAN or Near-RT RIC, as applicable for the use case. #. Non-RT RIC shall support relevant AI/ML model training based on the data in 1. for generating/optimizing policies and intents to guide the behavior of applications in Near-RT RIC or RAN, as applicable for the use case. #. Non-RT RIC shall support training of relevant AI/ML models based on the data in 1. to be deployed/updated in Near-RT RIC as required by the applications. #. Non-RT RIC shall support performance monitoring and evaluation. #. Non-RT RIC shall support a fallback mechanism to prevent drastic degradation/fluctuation of performance, e.g. to restore to the previous policy or configuration. The non-functional requirements are the following ones: #. Non-RT RIC shall not update the same policy or configuration parameter for a given Near-RT RIC or RAN function more often than once per second. #. Non-RT RIC shall be able to update policies in several Near-RT RICs. Moreover, there are functional requirements regarding the A1 interface: #. A1 interface shall support communication of policies/intents from Non-RT RIC to Near-RT RIC. #. A1 interface shall support AI/ML model deployment and update from Non-RT RIC to Near-RT RIC. #. A1 interface shall support communication of enrichment information from Non-RT RIC to Near-RT RIC. #. A1 interface shall support feedback from Near-RT RIC for monitoring AI/ML model performance. #. A1 interface shall support the policy/intents feedback from Near-RT RIC to Non-RT RIC.