About OAIC-T

OAIC-T is an open-source AI cellular testing platform which supports automated, distributed, and AI-enhanced testing of xApps in O-RAN.

Motivations of OAIC-T

While AI models are enablers to achieve intelligent next-G wireles networks, comprehensive testing of their performance is cumbersome and in many cases non-existent. This is mainly due to the inability of the current theory to explain or prevent failures in the AI models which are mainly trained in data-driven manners. Hence, it is necessary to have a framework and appropriate environment for testing AI models in their capacity of cellular RAN controllers. From the ongoing research and development and expected deployment of O-RAN components in Next-G networks, there is a urgent need for methods, platforms, and tools that facilitate testing various AI models in the radio network in a production like environment.

Key Design Requirements of OAIC-T

1. Software-defined and modular to enable customization;
2. Invasive/non-invasive testing during O-RAN operation in isolated or production environment to capture data in relevant operating conditions;
3. Open test interfaces to enable the development of new test methods and processes;
4. Test configuration files that enable specifying and reproducing a test;
5. Support for automated and AI-enhanced testing to assess the operation of AI-enabled cellular radio network controllers under a myriad of channel and contextual conditions (large search spaces)
6. Support for multitasking and distributed testing to enable a multi-user testing environment (e.g., producing different traffic scenarios)

OAIC-T Framework

An OAIC-T test involves automated setup of the testing environment, automated test execution, and automated generation of testing performance report. The OAIC-T framework consists of three major components: the OAIC-T server which sets up the testing environment as described in test configuration files and orchestrates the test execution as defined in test cases, the OAIC-T actor which executes test steps as instructed by the OAIC server, and the test repository which stores various test assets (e.g., test files, data files, log files, test results, etc.).

Major Release Milestones of OAIC-T

1. Phase I (Expected Date: Jan. 1, 2023) with implementations of core OAIC-T framework.
2. Phase II (Expected Date: June. 15, 2023) with implementations of AI testing methods, including Fuzzing, AI-Fuzzing, and adversarial learning.
3. Phase III (Expected Date: Dec. 15, 2023) with implementations of multitasking.
4. Phase IV (Expected Date: June. 15, 2024) with system integration and testing.

How to Run OAIC-T

Please refer to the OAIC-T document for details.