The dotted numbers look like a taxonomy. 802.1, 802.3, 802.11, 802.15: surely someone drew a diagram and assigned each branch a slot. They did not. The family tree is a record of arguments, and the first branch point was a failure to agree that got institutionalized instead of resolved.
The idea
IEEE 802 is not one standard with sections, it is a committee of committees with a constitutional layer. Each working group owns a medium and its access method, and 802.1 owns the architecture that all of them must remain compatible with. A new project anywhere in the family has to state, before it is authorized, whether it conforms to 802.1’s bridging model, and justify itself to 802.1 if it does not.
The split that made the numbering
The committee was chartered with a singular goal. “The IEEE 802 Local and Metropolitan Area Networks Committee was formed in 1980 to create a single standard for the lower layers of a local or metropolitan area network.” One standard. It failed at that in the most productive way available.
Three incompatible camps were in the room, each with a shipping or nearly shipping technology behind it, and “Because the group could not agree on a single approach, three working groups were formed, 802.3 for Ethernet (called CSMA/CD), 802.4 for token bus, and 802.5 for token ring.” The vendors could not be made to converge, so the committee converted an unresolvable technical disagreement into an organizational structure. That is the founding move of the entire family, and the reason 802 grows by adding working groups rather than by adding clauses.
The Ethernet branch had an existing document behind it: “In 1980, Digital, Intel, and Xerox published a standard called the DIX standard,” and “The first IEEE Standard for the CSMA/CD approach was based on the DIX standard.” A de facto multi-vendor specification walked into the process and came out with a committee’s name on it, which is the ordinary path for successful technology and the reason the ALOHA-descended access method of random-access radio ended up in an IEEE document at all.
Do not read too much into the number itself. “The number 802 has no significance: it was simply the next number in the sequence that the IEEE used for standards projects.”
The shape of the family
“The IEEE 802 family of standards has had twenty-four members, numbered 802.1 through 802.24, with a working group of the LMSC devoted to each,” and most of them are gone. 802.4 for token bus, 802.5 for token ring, 802.2 for logical link control, 802.6, 802.7, 802.12, 802.14, 802.17, 802.20 are disbanded. The family is a graveyard with a few very busy survivors: 802.1, 802.3, 802.11, 802.15, and a handful of advisory groups.
Disbanded is not the only terminal state. A working group can hibernate, and the operations manual specifies what that obliges: “The chair of a hibernating WG shall maintain a list of experts that are available to answer questions and provide clarification about the standards and/or recommended practices generated by the WG.” Someone must still be reachable to interpret a standard nobody is developing any more. That clause is a good measure of what a standards body actually sells, which is not documents but authoritative interpretation of documents.
Scope is bounded in ways worth knowing. “The IEEE 802 standards are restricted to computer networks carrying variable-size packets,” which is why cell relay and isochronous octet-stream networks were never 802 problems even though they are local networking.
The layering the family assumes
Every 802 standard is written against one architectural picture. “IEEE 802 divides the OSI data link layer into two sub-layers: logical link control (LLC) and medium access control (MAC),” and the MAC sublayer is where each working group’s individuality lives. Ethernet’s MAC and Wi-Fi’s MAC solve the same abstract problem (arbitrate a shared medium) with completely different mechanisms, and the split exists precisely so that everything above them can be common.
The ceiling is explicit: “Everything above LLC is explicitly out of scope for IEEE 802.” That single sentence explains why IEEE 802 and the IETF have coexisted for forty years without a jurisdictional war. IEEE owns layers one and two, the IETF owns three and up, and the seam between them is the LLC service interface. When you read the OSI model against the TCP/IP model, the 802 family is the part of OSI that survived as living engineering rather than as pedagogy.
What 802.1 does that the others do not
The interesting asymmetry is that one working group is not about a medium at all. 802.1 is the Higher Layer LAN Protocols Working Group, and it owns bridging, VLANs, security, and the overall architecture. It has no PHY.
Its authority is written into the project-authorization criteria the whole family runs on. Under compatibility, “Each proposed IEEE 802 LMSC standard should be in conformance with IEEE Std 802, IEEE 802.1AC, and IEEE 802.1Q,” and the enforcement clause is unambiguous: “If any variances in conformance emerge, they shall be thoroughly disclosed and reviewed with IEEE 802.1 WG prior to submitting a PAR to the Sponsor.”
Read that as a review gate. Before any working group in the family may even begin a project, it must confirm that its output will still bridge, still tag, and still fit the media-independent service model, and if it will not, it has to go and argue with 802.1 first. That is why a frame from 802.11 and a frame from 802.3 can meet inside one bridge, and why VLAN tagging is a single mechanism across radically different media rather than one invention per PHY.
The other gates on a new project
Compatibility is one of five criteria a proposed 802 standard must satisfy, and the rest are unusually candid about what a standards body is for. A project “shall have broad market potential,” it “shall provide evidence of a distinct identity,” and it must show technical and economic feasibility, with the economic case addressing installation and operational costs including energy consumption. Distinct identity is the anti-duplication rule, and it is the reason competing proposals get merged into one project rather than published as rivals. The 1980 split was the counterexample, and the family has spent four decades trying not to repeat it.
Wireless projects carry an additional obligation: “A WG proposing a wireless project shall demonstrate coexistence through the preparation of a Coexistence Assurance (CA) document unless it is not applicable.” Spectrum is shared, so a new radio standard must show it will not ruin the ones already there. That is a genuinely unusual requirement for a standards body to impose, and it exists because the medium itself is a commons.
Related Notes
- IEEE 802.1Q, VLAN Tagging for the 802.1 document every other working group must remain compatible with.
- IEEE 802.3, Ethernet as a Document for the branch that came from the DIX standard.
- IEEE 802.11, Wireless LAN as a Document for the branch whose MAC solves the same problem on a medium that cannot detect collisions.
- OSI and TCP/IP Models for the layering 802 assumes and the ceiling it refuses to cross.
- ALOHAnet and Random Access for the access method that became CSMA and then 802.3.
- QoS and Traffic Shaping for the service model that 802.1 carries across every medium in the family.
Sources
- IEEE 802 (Wikipedia) backs the twenty-four numbered members and their working groups, the variable-size-packet scope restriction, the LLC and MAC sublayer division, the out-of-scope ceiling above LLC, and the origin of the number 802.
- IEEE 802.3 (Wikipedia) backs the 1980 formation of the committee, the three-way split into 802.3, 802.4 and 802.5, and the DIX standard as the basis of the first CSMA/CD standard.
- IEEE 802 LMSC Operations Manual v19 backs the hibernating-working-group obligation, the compatibility criterion naming 802.1AC and 802.1Q, the 802.1 review requirement for variances, the broad-market-potential and distinct-identity criteria, and the coexistence assurance requirement for wireless projects.