Communication-Efficient and Distributed Learning Over Wireless Networks: Principles and Applications

Machine learning (ML) is a promising enabler for the fifth generation (5G)\ncommunication systems and beyond. By imbuing intelligence into the network\nedge, edge nodes can proactively carry out decision-making, and thereby react\nto local environmental changes and disturbances while experiencing zero\ncommunication latency. To achieve this goal, it is essential to cater for high\nML inference accuracy at scale under time-varying channel and network dynamics,\nby continuously exchanging fresh data and ML model updates in a distributed\nway. Taming this new kind of data traffic boils down to improving the\ncommunication efficiency of distributed learning by optimizing communication\npayload types, transmission techniques, and scheduling, as well as ML\narchitectures, algorithms, and data processing methods. To this end, this\narticle aims to provide a holistic overview of relevant communication and ML\nprinciples, and thereby present communication-efficient and distributed\nlearning frameworks with selected use cases.\n