Internet of Things Challenges and the Emerging Technology of TinyML
Vasileios Tsoukas, Anargyros Gkogkidis, Athanasios P. Kakarountas · 2023
The merger of Internet of Things (IoT) and Machine Learning undoubtedly offers several benefits, but due to the nature of IoT devices and their underlying design, a number of difficulties and challenges exist. The most fundamental of these has to do with the security dangers posed by the transmission of voluminous amounts of data, as well as several networking obstacles, such as delay and bandwidth constraints. Moving all processing close to the network's edge and preventing the transmission of data to third parties, such as the cloud or other devices, is a potential solution. Embedded Machine Learning, and TinyML in particular, is a new technology that allows the implementation of Machine Learning on hardware with limited resources, such as Microcontrollers and FPGAs. In this paper, the fundamental issues of IoT devices are examined, the TinyML technology's advantages are outlined, and a summary of the most recent TinyML apps and systems is offered.