M17 Identity and Addressing

There is no registry to join, and no number to apply for — not because nobody got around to building one, but because the protocol made one unnecessary. This is the story of how.

The Short Answer

To get on M17 you register nothing. You put your callsign in the radio and you transmit. There is no database, no application, no waiting for approval.

C4FM can say the same thing — but for an entirely different reason, and the difference is the whole point of this page.

That sounds like a slogan. It is not. It is a literal description of the frame, and it is worth understanding exactly how it works — because the engineering behind it is elegant, and once you see it you will understand why DMR could never have done the same thing.

Why DMR Needs a Registry and M17 Doesn’t

Start with the problem every digital voice protocol has to solve: when a transmission arrives, who is it from, and where is it going? That information has to travel inside the transmission itself, and it has to be small, because every bit spent on addressing is a bit not spent on voice.

DMR’s answer is a number. A DMR ID is a compact integer that fits comfortably in the frame — but a number is meaningless on its own. Something, somewhere, has to know that this number belongs to that operator. So a registry has to exist. Somebody has to issue the numbers, prevent collisions, and publish the lookup table. That is why RadioID.net exists: not as a policy decision, but as a mathematical consequence of routing on numbers.

D-STAR takes the opposite approach and sends the callsign itself — but it still requires you to register with a gateway before the network will carry your traffic.

M17 asked a different question: could you fit a real callsign into a field small enough to send in every single frame?

The answer turned out to be yes, and the margin is astonishingly tight.

Forty-Eight Bits, and Not One Bit More

An M17 address is 48 bits — six bytes. Into those six bytes goes up to nine characters of callsign.

Nine characters of ordinary text would need nine bytes in ASCII. To fit them into six, you have to stop encoding text and start encoding callsigns — and callsigns are a much smaller problem. They don’t contain lowercase letters. They don’t contain punctuation, or spaces, or accented characters. Strip the alphabet down to only what a callsign can actually contain, and each character needs far fewer bits.

This is base-40 encoding. There are exactly forty legal symbols, each character becomes a number from 0 to 39, and the whole callsign is composed into a single large integer — nine digits in base 40, rather than nine bytes.

Why forty? Because forty-one would not fit.

This is the detail that shows the design was derived rather than guessed. The specification works through the arithmetic explicitly:

What the forty symbols are — and what the spare ones bought

The bare minimum alphabet a callsign needs is thirty-seven symbols: twenty-six letters, ten numerals, and a forward slash. That leaves three slots spare inside the forty. They were not wasted.

The hyphen and the slash mean different things — deliberately

This is a genuinely thoughtful piece of design, and it is easy to miss.

• The hyphen creates a different station. AB1CD-1 and AB1CD-2 are separate stations as far as the protocol is concerned — while remaining plainly the same operator to any human reading them. This is the AX.25 SSID idea, carried into M17 intact. Your handheld and your hotspot can each have their own address without either of you needing a second callsign.
• The slash does not. AB1CD/M for mobile, or AB1CD/AE to advertise Extra-class privileges, are the same station — a temporary annotation of status, not a distinct entity.

One symbol partitions your identity into multiple stations. The other decorates it. The protocol knows the difference, and it was written down on purpose.

The Address Space

Forty-eight bits gives more addresses than the callsign alphabet can actually use, and the specification carves the remainder up rather than leaving it undefined:

That reserved block exists because forty raised to the ninth power does not quite reach two to the forty-eighth. It is the natural consequence of the encoding — the leftover space in a design that is otherwise pressed right up against its limit.

The Destination Doesn’t Have to Be a Person

Here is where the scheme earns its keep, and where it goes beyond anything the other modes do with identity.

The source address, in nearly all circumstances, decodes to an amateur callsign — that is you, and it rides in every frame. But the destination frequently is not a callsign at all. The same 48-bit field carries, without any special casing:

• Another station. A callsign — point-to-point, the obvious case.
• Everyone. The broadcast address.
• A reflector and module. The network destination, encoded exactly the same way as a callsign.
• A command. Words like ECHO or UNLINK — instructions to the infrastructure, in the destination field, because they encode in base-40 like anything else.

How It Compares

Read the last column. Only M17’s answer is a structural one. The others avoid a registry by policy, by circumstance, or not at all. M17 avoids one because the protocol left no job for it.

The Honest Trade-Off

The verification point is worth being plain about, because it is the standard objection. The answer is that no amateur mode meaningfully verifies identity on the air — a DMR ID proves someone registered a callsign once, not that the person keying the radio is that person. Amateur radio has always run on the honour system and self-identification. M17 simply declines to build machinery that would imply otherwise.

Quick Reference