MassiveIoTConnections

Abstract

By 2020, more than 50 billion internet of things (IoT) devices, ranging from critical IoT applications such as autonomous vehicles, manufacturing automation, and VR/AR to massive IoT applications such as smart home, smart farming, and smart energy, will be connected through radio communications. Cost‐effective connectivity and efficient management of such a huge number of heterogeneous devices are the main enablers for IoT services uptake. 5G is particularly suitable for IoT due to its key features as follows: (i) the integration of heterogeneous low‐power wide area (LPWA) radio access technologies, (ii) the introduction of fog/edge computing that brings the cloud functionalities close to front‐end devices, and (iii) the assistance of machine learning (ML) techniques that allows 5G wireless networks to be predictive and proactive. LPWA technologies can provide low‐rate long‐range radio communications, as a complement to the current cellular connectivity technologies. With the assistance fog/edge computing, the IoT network can scale more easily, and also end users can access the network resources more efficiently. ML helps an IoT network to make smart decisions via learning network uncertainties, planning resource allocations, and configuring the associated network parameters in a constantly varying environment. This article provides a comprehensive survey of these technologies. Specifically, the basic requirements and challenges in an IoT network with massive connections are reviewed. The advantages of using fog/edge‐based and ML‐based approaches are summarized. Furthermore, some potential issues caused by massive IoT connections for future work are also discussed.