Pinching Antennas Meet AI in Next-Generation Wireless Networks

TL;DR

NG wireless networks demand ultra-reliable, low-latency links in dynamic, blockage-prone environments. The paper proposes a PA–AI 'win-win' framework that uses pinching antennas along dielectric waveguides to create on-demand LoS links and employs AI to optimally deploy PA activations and manage resources, while PA-enabled channels bolster edge AI tasks like FL and AirComp. It outlines PA fundamentals, suitable ML tools (MLP, DQN, DDPG, GNN, MADQN, MADDPG), and PA-enabled edge AI scenarios, and discusses future directions including LLM-driven PA control and joint semantic communications/isac frameworks. The work demonstrates a path toward self-optimizing NG networks that combine low-cost, reconfigurable physical-layer methods with data-driven intelligent control to achieve robust, low-latency connectivity for emerging applications.

Abstract

Next-generation (NG) wireless networks must embrace innate intelligence in support of demanding emerging applications, such as extended reality and autonomous systems, under ultra-reliable and low-latency requirements. Pinching antennas (PAs), a new flexible low-cost technology, can create line-of-sight links by dynamically activating small dielectric pinches along a waveguide on demand. As a compelling complement, artificial intelligence (AI) offers the intelligence needed to manage the complex control of PA activation positions and resource allocation in these dynamic environments. This article explores the "win-win" cooperation between AI and PAs: AI facilitates the adaptive optimization of PA activation positions along the waveguide, while PAs support edge AI tasks such as federated learning and over-the-air aggregation. We also discuss promising research directions including large language model-driven PA control frameworks, and how PA-AI integration can advance semantic communications, and integrated sensing and communication. This synergy paves the way for adaptive, resilient, and self-optimizing NG networks.

Figures (4)

• Figure 1: The interplay between PA and AI.
• Figure 2: Overview of ML approaches for some PA-aided scenarios: centralized (MLP, DQN, DDPG), and distributed/multi-agent (GNN, MA-DQN, MA-DDPG), covering discrete/continuous activation positions, multi-waveguide coordination.
• Figure 3: PAs for edge AI networks: ① PAs for indoor FL network; ② PAs for AirComp-enabled model update; ③ PAs for real-time model training with data hotspot; ④ PAs for real-time model training with mobile devices.
• Figure 4: Performance of PA-assisted FL system.