Federated Learning for Internet of Things: Recent Advances, Taxonomy,\n and Open Challenges

The Internet of Things (IoT) will be ripe for the deployment of novel machine\nlearning algorithms for both network and application management. However, given\nthe presence of massively distributed and private datasets, it is challenging\nto use classical centralized learning algorithms in the IoT. To overcome this\nchallenge, federated learning can be a promising solution that enables\non-device machine learning without the need to migrate the private end-user\ndata to a central cloud. In federated learning, only learning model updates are\ntransferred between end-devices and the aggregation server. Although federated\nlearning can offer better privacy preservation than centralized machine\nlearning, it has still privacy concerns. In this paper, first, we present the\nrecent advances of federated learning towards enabling federated\nlearning-powered IoT applications. A set of metrics such as sparsification,\nrobustness, quantization, scalability, security, and privacy, is delineated in\norder to rigorously evaluate the recent advances. Second, we devise a taxonomy\nfor federated learning over IoT networks. Third, we propose two IoT use cases\nof dispersed federated learning that can offer better privacy preservation than\nfederated learning. Finally, we present several open research challenges with\ntheir possible solutions.\n