networks//IoT

The IoT (Internet of Things) is the practice of connecting sensors, machines and everyday devices to networks so that their data reach other systems (a dashboard, a database, a model in the cloud) and, sometimes, so that they can be commanded remotely; it is what lets a utility read ten thousand meters without sending anyone, a fleet operator see every drone's battery, or a plant gather the vibration of two hundred pumps into one place. Its typical device is small, cheap and battery or mains powered, an MCU with a radio (the ESP32 is the classic hobby and product part), publishing readings over MQTT or a similar lightweight protocol.

The IoT (Internet of Things) is the practice of connecting sensors, machines and everyday devices to networks so that their data reach other systems (a dashboard, a database, a model in the cloud) and, sometimes, so that they can be commanded remotely; it is what lets a utility read ten thousand meters without sending anyone, a fleet operator see every drone's battery, or a plant gather the vibration of two hundred pumps into one place. Its typical device is small, cheap and battery or mains powered, an MCU with a radio (the ESP32 is the classic hobby and product part), publishing readings over MQTT or a similar lightweight protocol.

What it brings is volume and reach; what it costs is latency, lost packets, storage and a larger attack surface. The book writes the cost as the time of one trip round any loop that crosses the network:

Tloop=Tmeasure+Tnetwork+Tcompute+Tact.T_{\text{loop}}=T_{\text{measure}}+T_{\text{network}}+T_{\text{compute}}+T_{\text{act}} .Tloop​=Tmeasure​+Tnetwork​+Tcompute​+Tact​.

If the network adds 200 ms with occasional seconds of delay, a loop closed through it cannot control anything that changes in less than a few seconds (loop delay). So fast control stays beside the machine, on the PLC or the microcontroller (edge computing), and the cloud supervises, stores, trains models and plans.

IoT connects and collects; a cyber-physical system closes the loop.

Two hundred pumps reporting a vibration summary every hour are IoT; the drive that trips one of them when its vibration crosses a limit belongs to a cyber-physical system, with timing, stability and physical safety as requirements. The two meet when IoT data feed slow decisions (a maintenance order, a setpoint changed once a day) and keep apart when a loop must answer in milliseconds.

Its industrial branch, the IIoT, uses the same ideas under harder demands: reliability, availability, security and integration with equipment that will not be replaced, carried by the plant's industrial data layer from edge gateway to historian.

It pays when there are many machines, remote sites, or a need for history to learn from. If an operator reads a gauge once a shift and that is enough to decide, a network adds cost and risk and nothing else.

Its typical failures are those of distributed systems: devices that go silent without anyone noticing (a missing reading is not an alarm unless something checks for it), clocks that disagree, firmware that cannot be updated in the field (OTA update), and default passwords on devices exposed to the internet, the commonest door for attackers.