In no doubt, IoT is network oriented - even the name IoT naturally relates to the Internet network.
IoT networking is much different than typical, multilayered, stack-oriented TCP/IP or similar communication model, we know well while using our PC, MAC, server or Smartphone on the daily basis.
Indeed constrained IoT devices are usually unable to operate regular - full time on, ISO/OSI layered stack, because of constrained resources. In details it primary means, IoT devices are limited by the processor power, RAM and storage sizes and particularly because of limited power resources. IoT device is expected to be energy efficient, thus low powered, that in most cases excludes typical wireless connection standards like i.e. WiFi. On the other hand, IoT devices are expected to communicate over long distances - some couple or a dozen of kilometres - where wired infrastructure like Ethernet cables and related infrastructure is non-existent and most of the wired technologies, copper-based is out of range.
Also, IoT devices daily life-cycle is much different than i.e. or PC life-cycle. We as humans used to switch on the notebook, work extensively on the web, then put it to the low power or off, making the machine to sleep, hibernate or simply shutting it down. They barely make network operations then. IoT devices are expected to be sleep and awake simultaneously almost all time and usually, they're expected to be full time operable as needed. Most tasks are periodical, i.e. working as a sensor-network node whereas period can be something like between seconds and months or even longer. Anyway, they're usually expected to trigger themselves to be awake from sleep, perform some operation and connect to the network.

Following chapters explain some most popular concepts about how to organize network fulfilling the above constraints on communication between IoT devices (Machine-2-Machine) and how to let them communicate with the Internet: including hardware, software and human-users.