Wireless communications — study guide

Cutting the last wire

A robot that stays on a bench can keep its cable. A robot that drives away cannot. The moment a machine has to be commanded and watched while it moves out of arm's reach, the wire that carried its steering commands and sent back its readings has to go, and something has to take its place. That something is a radio link: the same data, now riding through the air instead of down a conductor.

Trading a wire for the air is not a free swap. A wire gave two devices a private, dedicated path that was there whenever they wanted it. Radio gives them a shared stretch of open space that every other transmitter nearby is also using, that anyone in range can listen to, and that drops and garbles messages in ways a short clean wire never did. Every one of those changes has to be handled, and handling them is most of what makes a wireless link different from a serial cable.

This concept is a survey, not a teardown of any one radio. It lays out what changes when the medium becomes shared air, and then walks the common options a robot actually chooses from: Wi-Fi, Bluetooth, and simple long-range radio. The goal is that when a build needs to talk to something it cannot touch, you already know which link fits the job and what each one costs you to use.

From a private wire to open air

A wire is a private path. When two devices are joined by a conductor, the signal one puts on the line is the signal the other reads, and nothing else touches it. The connection is a physical fact: the copper is either there or it is not, and while it is there the two ends have the line entirely to themselves. Every wired link, including a plain serial connection between two chips, works this way - one path, two owners.

Radio keeps none of that. A transmitter does not reach out to one receiver; it turns its message into a wave and lets that wave spread outward into whatever space is around it. There is no conductor and no dedicated path - just a patch of air, and a receiver that happens to be close enough and tuned to the right frequency to pick the wave back up. It works like the difference between two cans joined by a taut string, where only the two ends share the sound, and calling out across an open room, where the sound reaches everyone present and carries only as far as your voice does.

Two consequences fall straight out of that picture, and the rest of this concept is mostly them. First, the signal reaches everyone in range, not just the intended receiver, so the link is defined by coverage and agreement rather than by a cord you can trace. Second, because the air belongs to no one, every other transmitter nearby is spreading its own waves through the same space at the same time. A wire handed you a clean, private channel for free; radio hands you an open, crowded one, and asks you to build the rest yourself.

Everyone talking on the same band

Radio does not use all of the air at once; it uses a slice of it called a band, a range of frequencies set aside for a purpose. The catch is that the most convenient bands are unlicensed and open to everyone, so they are crowded. The 2.4 GHz band is the busiest example: Wi-Fi uses it, Bluetooth uses it, and so do countless other gadgets and even microwave ovens, all radiating into the same frequencies in the same space at the same time.

When two transmitters on the same band talk at the same instant, their waves overlap at the receiver and blur into noise it cannot read, the way two people speaking over each other produce sound but no message. So devices sharing a band have to take turns: listen first, transmit when the channel seems clear, and back off and retry when a collision happens anyway. This is contention, and it is like a crowded room where everyone is on one channel: only one voice comes through clearly at a time, and when several talk at once their words overlap into noise nobody can read.

For a robot this is a running tax, not a one-time setup cost. The link works fine in a quiet room and then stumbles the moment the space fills with other radios - a competition venue full of teams, an office thick with Wi-Fi, a warehouse of machinery. Nothing broke; the band simply got crowded, and a crowded band means less airtime for every device on it, more collisions, and a link that slows down or drops out exactly when the surroundings get busy.

Range, data rate, and power: pick two

Three things a wireless link wants are how far it reaches, how much data it moves per second, and how little power it burns doing it. On a wire you rarely had to choose between them. On radio they pull against each other, and no link gets to maximize all three at once - a build picks which to favor and pays in the others. It works like choosing between a heavy truck that hauls a big load but only across town and burns fuel, and a lone runner who carries one short note for miles on a granola bar: you trade how much you carry against how far it goes and how much energy it costs.

The physics behind the pull is worth carrying as intuition. Moving more data per second needs a wider slice of spectrum and a stronger, more finely-read signal, which costs power and does not travel as far before it fades into the noise. Going farther on a fixed amount of power means slowing down - sending fewer bits per second so each one is easier to pick out at a distance. Lower frequencies also reach farther and pass through walls better than higher ones, which is part of why long-range links live down in the sub-gigahertz bands (around 868 MHz in Europe and 915 MHz in North America, as of 2026) while high-rate links sit up at 2.4 GHz and 5 GHz.

That is why the common wireless options land at such different points. A high-bandwidth link streams a video feed but reaches only tens of meters and drains a battery; a long-range link crosses a field on a coin cell but trickles only a few kilobits per second. Neither is better - they are answers to different questions, and a robot that needs both a fast nearby link and a slow far one simply carries two radios.

Delivery is no longer guaranteed

A short wire delivered nearly every bit it was handed. Radio does not. A wireless signal fades as it travels, reflects off walls into distorted copies of itself, and collides with other transmitters on the band, so some messages arrive garbled and some never arrive at all. Packet loss is not a fault on a wireless link; it is the normal weather, and the link is built to expect it.

The fix is to stop trusting a single send. The receiver acknowledges what it actually got, and the sender re-sends anything that goes unacknowledged, so a message that is lost the first time gets another chance rather than vanishing. That recovers reliability, but it buys it with time: every retry is a delay, and because losses come and go with the radio conditions, the delay is not steady. Latency on a wireless link jitters in a way a wire's did not, which matters most for the parts of a robot that need a command to land promptly and predictably.

None of this replaces the layered model - it fits inside it. Wireless is only the bottom of the stack, the physical and link layers that move raw bits through the air; the same higher-layer protocols a wired connection would use ride on top unchanged, including the reliable, ordered delivery that a transport protocol like TCP provides. The radio changes how bits cross the gap and how often they are lost; it does not change what the layers above it are for. That separation is the whole point of layering, and it is why a robot's software can often treat a flaky radio much like it treats a network cable.

Anyone in range can listen

To read the traffic on a wire, you had to reach the wire - open the case, clip onto the conductor, be physically present at the link. That physical barrier did a lot of quiet security work. Radio removes it completely. The signal spreads to every receiver in range whether or not it was meant for them, so anyone nearby with a matching radio can pick up a robot's traffic without touching a thing. It works like the difference between a postcard passed around a crowded room, which anyone who handles it can read and could scribble a fake reply onto, and a sealed envelope that hides its contents from every hand it passes through.

Listening is only half of it. A medium that anyone can hear is also a medium anyone can speak into. An attacker in range can transmit on the same band and try to inject their own commands, or drown the link in noise so the real controller cannot get through. A robot that took its wired link's privacy for granted and carried that assumption onto radio is a robot whose telemetry can be read and whose steering can be spoofed by a stranger across the parking lot.

So on radio, security cannot come from the medium; it has to be added on top. The link encrypts its traffic so that a listener in range hears only scrambled bits, and it authenticates its endpoints so that the robot ignores commands that do not come from its real controller. This is the same idea layering already gives you - protections carried by the higher layers - applied because the physical link they ride on is now open to everyone. Treat any unencrypted wireless link as a public broadcast, because that is exactly what it is.

Wi-Fi, Bluetooth, and plain radio

Three families of wireless link cover most of what a robot needs, and each one sits at a different point in the range, rate, and power tradeoff. Knowing where each lands is most of knowing which to reach for.

Wi-Fi is the high-bandwidth option. It moves a lot of data - tens to hundreds of megabits per second in practice, as of 2026 - over roughly tens of meters indoors, on the 2.4 GHz and 5 GHz bands. It also draws the most power of the three, far more than a Bluetooth radio, so it wants a real battery rather than a coin cell. That profile fits the data-heavy jobs on a robot: streaming a camera feed, carrying rich telemetry, or letting a laptop talk to the robot's onboard computer.

Bluetooth, and especially its low-energy variant BLE, trades bandwidth for frugality. It carries only a megabit or two per second over a short range - on the order of ten meters - but sips so little power that a BLE sensor can run for years on a small battery. It suits the light, occasional jobs: pushing a configuration from a phone, reading a handful of sensor values, waking a device up. Simple radio links take the opposite extreme from Wi-Fi. A plain sub-gigahertz radio, of the kind used for hobby RC control and for LoRa-style long-range links, reaches much farther - hundreds of meters to kilometers - on very little power, but only by trickling data, often just a few kilobits per second. That is exactly the budget a steering-and-throttle command needs and nothing more, which is why direct robot control so often rides a simple radio rather than Wi-Fi.

No single family wins, so real builds mix them: a fast link for the data flood, a frugal one for the trickle, a far one for control at a distance. Choosing among them is just reading a peripheral's demands against the same range, rate, and power axes every wireless link is measured on.

Driving a robot you cannot reach

A small self-driving car is the case that makes all of this concrete. It moves, which is the whole reason it exists, and the moment it moves it leaves the range of any cable. Yet it still has to be told where to go and still has to report what it sees, so every command out to it and every reading back from it crosses a radio link instead of a wire. Cutting that last wire is not an add-on for a mobile robot; it is the thing that lets it be mobile at all.

One radio rarely covers the whole job, because the car's links sit at different points in the tradeoff. A camera feed and detailed telemetry are a flood of data over a short hop, which is Wi-Fi's ground; the steering and throttle commands are a thin trickle that must arrive promptly and reliably even at a distance, which is a simple low-latency radio's ground; a quick configuration from a phone on the bench is BLE's. A serious build may carry more than one radio for exactly this reason, each picked by the same range, rate, and power questions that sort every wireless option.

What travels over those links is not new. A wireless connection carries the same kinds of serial and message data a wired bus between two chips would carry - the readings and commands are unchanged; only the medium beneath them became shared air. That is why this concept closes the hardware-communications thread: once you can move a robot's data over a wire and over the air, and choose sensibly between them, the machine is finally free to leave the bench.