Interactive AI with Retrieval-Augmented Generation for Next Generation Networking

TL;DR

The paper investigates interactive AI (IAI) as a pathway toward agile, context-aware networking in next-generation environments. It proposes an IAI-enabled network management framework featuring environment, perception, action, and brain units, augmented with a pluggable LLM and a retrieval-augmented memory (RAG) module implemented via LangChain. A case study on RIS-assisted SWIPT with RSMA demonstrates automatic problem formulation by the IAI agent, achieving results that closely match ground-truth formulations and outperform manual approaches. The work outlines future directions in integrating IAI with edge and semantic technologies, enhancing security, and establishing evaluation criteria for IAI-generated solutions, underscoring the practical impact of interactive, knowledge-backed AI in dynamic networks.

Abstract

With the advance of artificial intelligence (AI), the emergence of Google Gemini and OpenAI Q* marks the direction towards artificial general intelligence (AGI). To implement AGI, the concept of interactive AI (IAI) has been introduced, which can interactively understand and respond not only to human user input but also to dynamic system and network conditions. In this article, we explore an integration and enhancement of IAI in networking. We first comprehensively review recent developments and future perspectives of AI and then introduce the technology and components of IAI. We then explore the integration of IAI into the next-generation networks, focusing on how implicit and explicit interactions can enhance network functionality, improve user experience, and promote efficient network management. Subsequently, we propose an IAI-enabled network management and optimization framework, which consists of environment, perception, action, and brain units. We also design the pluggable large language model (LLM) module and retrieval augmented generation (RAG) module to build the knowledge base and contextual memory for decision-making in the brain unit. We demonstrate the effectiveness of the framework through case studies. Finally, we discuss potential research directions for IAI-based networks.

Figures (4)

• Figure 1: The evolution of AI. For IAI, we highlight the role of the mixture of experts, large language models, deep reinforcement learning, retrieval-augmented generation, and generative AI in promoting adaptability and interaction with users. Additionally, the components and advantages of IAI are also presented, emphasizing its efficiency and adaptability in different applications such as optimization and traffic management.
• Figure 2: In the IAI-enabled problem formulation framework, three key components are presented: brain, perception, and action. The brain units with LLM and RAG modules as the central processor, handling essential tasks such as memory retention, retrieval processing, and making decisions. Meanwhile, the perception module is responsible for acquiring and interpreting diverse data from the environment. Lastly, the action module implements responses and interacts with the environment, utilizing various tools for execution.
• Figure 3: The chunk size versus the number of interaction rounds required to solve the network optimization task, where $k$ represents the number of chunks that the LLM puts into context in each round of interaction.
• Figure 4: Comparison of system performance under various optimization problem generation methods. The figure displays the effectiveness of the proposed IAI framework modeling against traditional real modeling (upper bound) and manual modeling approaches, where the PPO-based DRL method is set as the solution method to demonstrate the performance results.