Beyond 6G: Multimodal Approaches for Joint Communication and Sensing

This position paper asserts that joint communication and sensing (JC&S) in 6G and beyond must adopt multimodal approaches to fully realize its potential. Traditional statistical or signal processing methods alone are insufficient to manage the complexity of 6G networks. Machine learning (ML) and artificial intelligence (AI) are essential, not only for real-time decision-making but also for driving large-scale simulations that generate synthetic data for model training. We propose using computer vision to annotate radio data, improving ML models, while AI-in-the-loop simulations operating in fivedimensional (5D) environments-integrating spatial, temporal, and spectral information-will continuously refine sensing and communication policies. This adaptive system will allow JC&S devices to dynamically switch between AI-based learning and real-time tasks. We explore the data, algorithmic, hardware, and application levels of this integration, and discuss practical implementations, challenges, and future directions for advancing next-generation JC&S systems.