The number of cellular IoT connections is set to grow by three billion globally in the next four years, according to a report by Juniper Research.

This will necessitate federated learning to automate IoT management.

The study found the global number of cellular IoT devices will increase from 3.4 billion in 2024 to 6.5 billion by 2028. However, the study predicts that this 90% growth in connections will require the deployment of new services enabling the efficient automation of IoT device management and security.

The research identified intelligent infrastructure management, which enable IoT users to automate the configuration of devices, security processes and connectivity in real time, as key to handling the large increase in cellular data.

The research firm anticipates that global cellular IoT data generated will grow to 46 petabytes in 2028, up from 21 petabytes this year, leading to further investment in IoT automation services, such as federated learning.

At present, most machine-learning models are trained via data sources that are stored in a single location, making opportunities for fraudulent players a simpler task. In response, the study recommends operators transition to federated learning models, a subset of machine learning that leverages a decentralised data approach to reduce the chances of data fraud over IoT networks.

Federated machine learning limits the exposure of sensitive IoT data, thus reducing the threat of data breaches.

“As the number of cellular IoT connections grows, it is imperative that both platforms and operators ensure data are secure in transition and on device,” said research author Alex Webb. “A failure to do so will dissuade IoT users in industries with sensitive data from using a cellular IoT-based approach to connectivity.”

An extract from the report “Global Cellular IoT Market: 2024-2028” is available as a free download here: www.juniperresearch.com/research/telco-connectivity/operator-strategies/cellular-iot-strategies-research-report.