Distributed Learning for 6G–IoT Networks: A Comprehensive Survey

• DOI: 10.36227/techrxiv.20069051.v2
• Publication Date: 2022-06-22T14:45:20Z
• Authors: Sree Krishna Das; Ratna Mudi; Md. Siddikur Rahman; Abraham O. Fapojuwo

<p>Smart services based on the Internet of Things (IoT) are likely to grow in popularity forthcoming years, necessitating improvement fifth-generation (5G) cellular networks upgrade future from their present state. Despite fact that 5G may manage a diversity IoT services, they not be able fully meet requirements emerging smart applications due limitations that, many cases, could overcome by applying artificial intelligence (AI). Therefore, sixth–generation (6G) wireless technologies being developed address networks. Traditional machine learning (ML) techniques driven centralized way. However, huge volume produced data, confidentiality concerns, and growing computing competencies edge devices have led exposure promising solution decentralized way which is called distributed learning. This paper provides comprehensive analysis (e.g., federated (FL), multi–agent reinforcement (MARL)–based FL framework) how deploy an effective efficient for Moreover, we describe timely review role facilitating 6G enabling technologies, such as mobile computing, network slicing, satellite communications, terahertz links, blockchain, semantic communications. Also, identify discuss several open research issues related FL–empowered In particular, focus extensive range applications. For each application, main motivation using along with associated challenges detailed examples use scenarios given. Regarding AI techniques, consider MARL–based framework tailored needs ensuring fast convergence high model accuracy large state action spaces. Particularly, varying radio channels limited resources transmission power spectrum) communication environment, this article proposes robust enable local users perform allocation, mode selection, resource interference management. Finally, outlines prospective upcoming topics, aimed create constructive incorporation networks.</p>