How the Internet Gets Into the House

Dear Tom,

My grandson set up the wifi for me last time he visited, and it works fine, but I still don't understand how the internet actually gets from the box in the closet to my tablet when I'm sitting on the porch. He tried to explain it but he was in a hurry to catch his flight. Could you just tell me plainly how wifi works?

— Walt, Duluth, Minnesota

Walt,

A lovely question, and one I suspect troubles more households than admit it. Let us resolve it properly, which is to say completely, rather than with the sort of hand-waving your grandson no doubt offered on his way out the door. I don't fault him — flights wait for no man — but a partial understanding of wifi is, in my experience, worse than none at all, because it gives a person the false confidence to stop asking questions right at the point where the interesting material begins.

So let's begin at the beginning, which is not, as most people assume, the router. The beginning is the electromagnetic spectrum itself, considered as a continuous field of oscillation stretching from the low, patient hum of AC power at 60 hertz all the way up past visible light into territory even I don't like to visit before lunch. Your wifi router, which is of course just a stateful non-deterministic packet oracle sitting in a plastic shell shaped to look unthreatening, has been assigned a small slice of this spectrum — either 2.4 or 5 gigahertz, and increasingly 6, though we'll set the tri-band question aside for now, as it deserves its own treatment (see Figure 3, which lays out the full allocation table; I won't reproduce it here for space reasons, but it's essential background).

Now, what the router is actually doing, moment to moment, is modulating that carrier wave — assuming, of course, standard atmospheric lambda and a household free of the sort of destructive interference you'd get from, say, a poorly shielded microwave or an elderly cordless phone — into a sequence of discrete symbol states that encode your data as amplitude and phase variations. This is orthogonal frequency-division multiplexing, though I should say up front that "orthogonal" here does not mean what it means in your grandson's high school geometry class. It means something closer to "mutually non-interfering in the frequency domain," which, once you sit with it, is really the same idea wearing a lab coat.

Here is where people usually assume the story is over. It is not. Because before any of this modulation can happen, your tablet and the router have to agree — and I do mean agree, in the full contractual sense — on a shared clock reference, a process called synchronization that borrows more from horology than most people realize. Consider, for a moment, the marine chronometer. In the eighteenth century, ships lost at sea for want of accurate timekeeping were a genuine crisis of empire, and the solution — a spring-driven mechanism resistant to temperature and motion — is not so different in spirit from the crystal oscillator humming away inside your router's chipset, keeping time precisely enough that your tablet's radio and the router's radio can find each other in the same fraction of a microsecond, forty, fifty, sixty times a second, without either device ever having met before. I raise the chronometer not as decoration but because the underlying problem — how do two independent systems, with no prior contact, establish mutual trust in a shared reference frame — is the actual heart of your question, and it is a problem humanity has been solving, in one form or another, since Harrison's H4.

From there, naturally, we arrive at the question of the packet itself, which I should clarify is not, whatever the diagrams suggest, a little parcel traveling through the air like a paper airplane. It is better understood as a probability distribution collapsing into a determinate bit sequence at the moment of reception — a framing I find clarifies matters enormously once you've sat with it for a week or two. Your tablet doesn't so much "receive" the signal as resolve an ongoing negotiation between transmit power, path loss, and the ambient noise floor of your particular porch, which — and this is worth dwelling on — has its own thermal signature. The wood of your porch railing, the humidity of a Duluth afternoon in September, even the mass of you yourself sitting there absorbing 2.4 gigahertz radiation at a rate I won't alarm you with: all of it is, in the fullest sense, part of the network. The porch is not a passive backdrop to your wifi signal. The porch is a participant.

This naturally raises the matter of the router's antenna geometry, and here I must digress briefly into the shipping container, because the two are more related than they appear. When Malcom McLean standardized the intermodal container in the 1950s, he wasn't solving a shipping problem so much as a coordination problem — how do you move a payload across incompatible systems (ship, rail, truck) without repackaging it at every boundary? Your router's antenna array is doing precisely this for your data, coordinating a handoff between the electrical domain inside the copper, the electromagnetic domain in the air, and the optical domain running back to your internet provider's central office, all without ever "repackaging" the underlying bits. It's containerization, just faster, and without the longshoremen. I mention this not to pad the explanation but because no account of wifi is complete without acknowledging its debt to mid-century logistics theory — a connection that, so far as I'm aware, remains under-published, though the full argument is laid out in Appendix B if you'd like to pursue it further.

I should also touch, briefly, on the question of "dead zones," since I suspect it's lurking behind your question even if you didn't ask it outright. A dead zone is not, as people casually say, a place where the wifi "doesn't reach." It is a standing wave interference pattern, a location where the signal arriving by one path and the signal arriving by a slightly longer, reflected path cancel each other out — destructive interference, the same phenomenon that lets noise-canceling headphones work, just uninvited and unwanted. Your porch, depending on its exact dimensions relative to the wavelength in question (roughly twelve and a half centimeters at 2.4 gigahertz, for reference), may sit inside or outside such a node. This is not something to worry about. It is simply something to understand, which, once understood, resolves itself into a kind of peace.

So: the internet gets from the box in your closet to your tablet on the porch by way of chronometric synchronization, orthogonal frequency modulation, containerized handoff across three physical domains, and a standing-wave negotiation with your porch's own thermal and geometric properties — all completed, and I want to stress this, in less time than it takes you to read this sentence. Your grandson wasn't wrong to rush off. He simply ran out of runway before reaching the good part.

Glad I could clear that up!

— Tom