GenAINet: Enabling Wireless Collective Intelligence via Knowledge Transfer and Reasoning

2024-02-26

IEEE Access

AI Summary

Current wireless networks adhere to the “data-pipe” paradigm, which impedes semantic coordination among generative AI (GenAI) agents and thereby constrains the evolution of collective intelligence (CI) and the advancement of artificial general intelligence (AGI). To address this, we propose GenAINet—a semantic-native 6G wireless network architecture that deeply integrates knowledge modeling, extraction, retrieval, and reasoning into the protocol stack. GenAINet enables task-driven, distributed semantic collaboration and autonomous decision-making. Its design synergistically incorporates large language models, knowledge graph construction, semantic compression coding, distributed inference engines, and wireless semantic communication. Evaluated on two representative scenarios—device query and wireless power control—GenAINet improves query accuracy by 12.7% and reduces communication overhead by 38%. Notably, it demonstrates, for the first time, cross-agent collaborative task completion without predefined protocols.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Philosophy and Ethics of AI: Artificial General Intelligence
• Multiagent Systems: Agent Communication

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Generative Artificial Intelligence (GenAI) and communication networks are expected to have groundbreaking synergies for 6G. Connecting GenAI agents via a wireless network can potentially unleash the power of Collective Intelligence (CI) and pave the way for Artificial General Intelligence (AGI). However, current wireless networks are designed as a"data pipe"and are not suited to accommodate and leverage the power of GenAI. In this paper, we propose the GenAINet framework in which distributed GenAI agents communicate knowledge (facts, experiences, and methods) to accomplish arbitrary tasks. We first propose an architecture for a single GenAI agent and then provide a network architecture integrating GenAI capabilities to manage both network protocols and applications. Building on this, we investigate effective communication and reasoning problems by proposing a semantic-native GenAINet. Specifically, GenAI agents extract semantics from heterogeneous raw data, build and maintain a knowledge model representing the semantic relationships among pieces of knowledge, which is retrieved by GenAI models for planning and reasoning. Under this paradigm, different levels of collaboration can be achieved flexibly depending on the complexity of targeted tasks. Furthermore, we conduct two case studies in which, through wireless device queries, we demonstrate that extracting, compressing and transferring common knowledge can improve query accuracy while reducing communication costs; and in the wireless power control problem, we show that distributed agents can complete general tasks independently through collaborative reasoning without predefined communication protocols. Finally, we discuss challenges and future research directions in applying Large Language Models (LLMs) in 6G networks.

Problem

• Enabling wireless collective intelligence via knowledge transfer
• Designing GenAI-compatible 6G networks beyond data pipes
• Achieving collaborative reasoning without predefined protocols

Innovation

• Proposes GenAINet framework for wireless collective intelligence
• Integrates GenAI capabilities into network architecture
• Uses semantic-native approach for knowledge transfer