A Brief Review of Machine Learning-Based Approaches for Advanced Interference Management in 6G In-X Sub-networks

6G is envisioned to take the form of a ‘network of networks’, where wireless networks of different types and operational ranges are to be seamlessly integrated. In-X sub-networks are short-range low power radio cells to be installed at the very edge of such “network of networks” inside of different entities like a production module or a vehicle, and supporting highly localized yet different services with extreme requirements. Though benefiting from decision-making capabilities of the umbrella 6G network when available, these sub-networks are by default independent mobile cells sharing the same spectrum and thus leading to dense scenarios with severe and time-varying interference that should be carefully handled. However, managing such interference is an intractable challenge to traditional approaches. Therefore, driven by the benefits and efficiency of Machine-Learning (ML), this article investigates ML-based approaches, including supervised, unsupervised, and reinforcement learning approaches for advanced interference management in 6G in-X sub-networks. To this end, we overview the proposed schemes in the literature and discuss their methodologies, strengths and weaknesses. Furthermore, we highlight related open issues and future research directions, and we propose some useful guidelines for an ML-based solution.