Tactile internet of federated things: Toward fine-grained design of FL-based architecture to meet TIoT demands

Abstract

The Tactile Internet of Things (TIoT) represents a special class of the Internet of Things (IoT) that has opened the door for a new generation of agile, highly dynamic, intelligent, and bandwidth-hungry applications in many vital domains such as medical, military, robotics, and entertainment. However, the TIoT’s high level of physical requirements constitute a challenge that hinders the technology’s full potential. The application of machine learning (ML) methods has become of high importance to facilitate the prolific deployment of TIoT devices and applications. Along with advances in 6G and networking-enabling technologies, federated learning (FL) can bridge this gap by boosting TIoT services due to its inherent and redundant distributed nature. To the best of our knowledge, no other work in the literature covers applications of FL in the TIoT. In this work, we identify the prospective approach with a final aim of finding an appropriate architecture for the TIoT based on FL as we survey, compare, and analyze recent literature. We conducted a comprehensive and in-depth study of FL design aspects relevant to the deployment of the TIoT and suggested categorization of these aspects into process, infrastructure, data, enabler, and defense classes. On the other hand, as we are seeking to define a role for FL in the TIoT, we identified services of ML in the IoT domain as a general case of the FL/TIoT special case. Moreover, the shortcomings of traditional ML in regard to the general IoT are highlighted. We find that FL can be replaced, especially in the case of the TIoT, to meet the high level of design constraints. Eventually, we draw conclusions and future directions of common challenges relevant to applying FL in the domain of the TIoT.