Wireless Large AI Model: Shaping the AI-Native Future of 6G and Beyond

TL;DR

This survey defines wireless large AI models (WLAM) as foundation-scale AI systems embedded in 6G and beyond, outlining a dual, synergistic relationship: AI-native networks leverage WLAMs as network brains, while wireless infrastructure enables scalable AI training and inference. It presents a comprehensive taxonomy of WLAM fundamentals (architectures, data pipelines, RL, fine-tuning, inference, multimodal alignment, and deployment) and surveys WLAM applications across physical, network, and semantic layers, including wireless agents and ISAC. The work further discusses how wireless technologies (edge intelligence, FL/SL/FSL, AirComp, and PLS) enable efficient WLAM operation, and examines emerging technologies (PINNs, hypernetworks, next-gen sequence models, HDC, quantum ML) that could shape future WLAMs. Finally, it identifies high-level challenges—datasets, standardization, energy efficiency, security/privacy, and channel effects on training—and outlines future directions toward edge-native AI, trustworthy autonomous WLAMs, and physics-informed integration. The paper argues that achieving AI-native 6G hinges on holistic, cross-layer design that unifies advanced ML with robust wireless infrastructure. Its findings underscore the practical significance of WLAM in achieving automated, scalable, and semantically-aware wireless networks that can adapt to dynamic environments and diverse services.

Abstract

The emergence of sixth-generation and beyond communication systems is expected to fundamentally transform digital experiences through introducing unparalleled levels of intelligence, efficiency, and connectivity. A promising technology poised to enable this revolutionary vision is the wireless large AI model (WLAM), characterized by its exceptional capabilities in data processing, inference, and decision-making. In light of these remarkable capabilities, this paper provides a comprehensive survey of WLAM, elucidating its fundamental principles, diverse applications, critical challenges, and future research opportunities. We begin by introducing the background of WLAM and analyzing the key synergies with wireless networks, emphasizing the mutual benefits. Subsequently, we explore the foundational characteristics of WLAM, delving into their unique relevance in wireless environments. Then, the role of WLAM in optimizing wireless communication systems across various use cases and the reciprocal benefits are systematically investigated. Furthermore, we discuss the integration of WLAM with emerging technologies, highlighting their potential to enable transformative capabilities and breakthroughs in wireless communication. Finally, we thoroughly examine the high-level challenges hindering the practical implementation of WLAM and discuss pivotal future research directions.

Figures (19)

• Figure 1: The synergistic relationship between Large AI for Wireless and Wireless for Large AI. The top half illustrates the top-down optimization where AI acts as the network brain, while the bottom half depicts the bottom-up infrastructure support provided by the wireless network for AI training and inference.
• Figure 2: The outline of this survey.
• Figure 3: The KPIs for WLAM.
• Figure 4: The outline of Section 2.
• Figure 5: Advanced reinforcement learning algorithms.