Multi-Agent-Based Simulation of Intelligent Network System
Petro Mushidi Tshakwanda, Sisay Tadesse Arzo, Michael Devetsikiotis · 2023
The agent-based approach to service design offers a distinct advantage over traditional microservice-based design by providing not only reactive responses but also the ability to proactively anticipate and address potential issues. This, coupled with other benefits, has made the use of multi-agent-based intelligent network service design a popular and cutting-edge paradigm in the field of network research. Recently, the introduction of a softwarized intelligent network architecture and design guidelines using agents as building blocks, has been proposed for the upcoming 6G networks. However, despite this progress, there remains a lack of suitable simulation environments for evaluating the performance of multi-agent systems within these softwarized networks. In this work, we present the ongoing development and implementation of various network service agents utilizing the PADE framework. To facilitate this, we have also provided a GitHub link for the project. The implementation makes use of various machine learning and deep learning algorithms as possible evaluation scenarios. This experimental environment can also be utilized to perform a wide range of intelligent algorithm experiments.