WirelessGPT: A Generative Pre-trained Multi-task Learning Framework for Wireless Communication

TL;DR

WirelessGPT introduces a Transformer-based foundation model pretrained on large-scale wireless channel data to learn universal, cross-domain representations for both communication and sensing tasks. By leveraging multi-dataset pretraining (Traciverse, SionnaRT, DeepMIMO), it supports diverse ISAC tasks with minimal fine-tuning, and demonstrates improvements in channel estimation, competitive channel prediction, high-accuracy activity recognition, and accurate environmental reconstruction. The framework emphasizes scalability (600K–800M parameters) and a pipeline from pretraining to universal representation generation and downstream task execution, enabling data-efficient, multi-task wireless intelligence. Overall, the work establishes a scalable, unified approach to ISAC and multi-task wireless systems, paving the way for future foundation-model-based wireless solutions.

Abstract

This paper introduces WirelessGPT, a pioneering foundation model specifically designed for multi-task learning in wireless communication and sensing. Specifically, WirelessGPT leverages large-scale wireless channel datasets for unsupervised pretraining and extracting universal channel representations, which captures complex spatiotemporal dependencies. In fact,this task-agnostic design adapts WirelessGPT seamlessly to a wide range of downstream tasks, using a unified representation with minimal fine-tuning. By unifying communication and sensing functionalities, WirelessGPT addresses the limitations of task-specific models, offering a scalable and efficient solution for integrated sensing and communication (ISAC). With an initial parameter size of around 80 million, WirelessGPT demonstrates significant improvements over conventional methods and smaller AI models, reducing reliance on large-scale labeled data. As the first foundation model capable of supporting diverse tasks across different domains, WirelessGPT establishes a new benchmark, paving the way for future advancements in multi-task wireless systems.

Figures (4)

• Figure 1: Framework of WirelessGPT
• Figure 2: Framework illustration of pretraining
• Figure 3: NMSE comparisons between WirelessGPT and baseline models for channel estimation
• Figure 4: NMSE comparisons against channel quality for channel prediction a) WirelessGPT vs Transformer/LSTM b) WirelessGPT vs large language model