Intent Manager Function with Actuation Predictions: A Performance Evaluation

The vision towards 6G incorporates new technologies enabling cognitive operations, which demands for enhanced automation due to complexity. Intent-driven networks need to support multiple intents from different sources simultaneously for enhanced automation. There are multiple reasons why conflict might occur among expectations formulated in intents, with some prominent examples being (1) autonomous operations, and (2) restricted capacity of network and computation resources. Predicting the effect of a configuration on the system is a promising approach to detect potential conflicts, but its efficiency highly depends on the prediction performance. In this study, we propose a performance evaluation framework for predictions, so that further improvements can be carried out to enhance the overall performance of the intent handling operations with integrated conflict detection and resolution capability. As a result of experiments with three diverse services, it is observed that the predictions we modeled can estimate the effect of a proposed action on system accurately, and so potential conflicts can be detected and resolved on the run.