Link 11 (TADIL A) migration: getting off the legacy data link

Link 11 — TADIL A in US parlance — is the tactical data link that connected NATO surface and air-defense forces for half a century. It still runs on ships and shore sites that were laid down before the data link community settled on TDMA. Most navies now treat it as a liability rather than a capability: the waveform is slow, the crypto is end-of-life, the spares are dwindling, and the operators who can troubleshoot a roll-call dropout are retiring faster than they are replaced. This is a migration playbook — what Link 11 actually does, why it has to go, and the two viable exit paths, with the engineering reality of running both old and new at once.

1. link 11 today

Link 11 is a half-duplex, netted data link governed by STANAG 5511 and MIL-STD-6011. It runs on two physical bearers: HF (2–30 MHz) for over-the-horizon range via skywave, and UHF (225–400 MHz) for line-of-sight. The defining feature is its access scheme. One participating unit is designated the Net Control Station (NCS), and the NCS runs a continuous roll-call: it interrogates each Picket Unit (PU) in turn by address, the addressed PU transmits its update, then control returns to the NCS to poll the next address. The cycle repeats forever. There is no contention and no time-slot map — just a serial, master-driven polling loop.

The payload is the M-series message catalog, also defined in MIL-STD-6011. M-series messages are 24-bit data words (with parity, 30 bits on the wire) carrying track position, identity, IFF, and management data — conceptually the ancestors of Link 16's J-series. The radios are driven by a Data Terminal Set (DTS), the modem that converts the conventional serial message stream to and from the audio-tone waveform the radio transmits. The classic DTS waveform packs 16 tones (the "fast" Link 11 mode) into a 1364 Hz audio band; the legacy "slow" mode is 75 baud. Either way the netted throughput is a few kilobits per second shared across the whole net.

2. why migrate

The roll-call architecture is the root problem. Because the NCS polls each PU serially, the latency to update any one track scales with the number of participants: a thirty-unit net means a track update can be a full polling cycle stale before it propagates. Half-duplex means no unit transmits while another is being polled, so the link cannot overlap traffic the way a TDMA waveform does. Add HF propagation, which fades, multipaths, and drops with the ionosphere and the time of day, and you have a link whose effective track-capacity and reliability degrade exactly when the operational picture is busiest.

The non-technical reasons are just as decisive. The crypto devices that key Link 11 are largely obsolete and increasingly unsupportable. DTS hardware spares come from a shrinking pool of legacy suppliers, and the institutional skill to align a Link 11 net — frequency, audio levels, NCS designation — is concentrated in a generation of operators leaving service. A link that cannot be repaired and cannot be staffed is not a capability you are keeping; it is a risk you are deferring.

3. the link 22 path

Link 22, formally NILE (NATO Improved Link Eleven), is the standard the alliance built specifically to replace Link 11. That lineage matters: it inherits Link 11's beyond-line-of-sight mission and dual HF/UHF bearer set, but replaces the master-driven roll-call with a dynamic TDMA architecture managed by a Network Controller. STANAG 5522 governs it. The message payload is the F-series — designed as a superset compatible with both M-series and J-series semantics, so an F2 maps onto a J2 and an M-series position report alike. The deeper trade-offs between the two next-gen links are covered in our Link 22 vs Link 16 analysis.

For a Link 11 operator the Link 22 path is the natural one: it preserves the over-the-horizon HF range that justified Link 11 in the first place while removing the polling bottleneck. The hardware story is the Network Interface Unit (NIU), the box that hosts the Link 22 protocol stack and drives the radios; NIU implementations come from a small set of qualified suppliers (Rockwell Collins, Thales, Leonardo), and installation on a combatant is a non-trivial integration, not a card swap. Lead times run in the years, which is the single biggest scheduling constraint on the whole migration.

4. the link 16 + JREAP path

If the operational requirement Link 11 was filling is really "share the air picture," the answer may be Link 16 rather than Link 22. Link 16 (STANAG 5516, MIL-STD-6016) carries the densest, lowest-latency air picture in the alliance, but it is line-of-sight UHF — it does not reach over the horizon the way Link 11 HF did. That gap is bridged by JREAP, the Joint Range Extension Application Protocol, which tunnels Link 16 J-series messages over non-Link-16 bearers: JREAP-A over satellite, JREAP-B over point-to-point serial, JREAP-C over IP.

The practical pattern is to retire Link 11 in favor of Link 16 inside the line-of-sight bubble and JREAP-C to extend that picture to dispersed nodes over IP/SATCOM. The cost is latency — a JREAP-C tunneled track update adds hundreds of milliseconds of network delay on top of the Link 16 slot schedule — and the need for gateway translation at every boundary where Link 11's M-series has to become J-series. That translation is exactly the kind of work a dedicated tactical data link gateway exists to do.

5. data forwarding and the DLP

During any migration the old and new links coexist, and the component that holds them together is the Data Link Processor (DLP) — sometimes the Forwarding unit in NATO terminology. The DLP correlates tracks across links, resolves duplicate reports of the same contact, and forwards data between Link 11, Link 16, and Link 22 so that a unit on the legacy net still sees tracks reported only on the new one and vice versa. It is the single point where the migration is made invisible to the operator.

Forwarding M-series to J-series is where the fidelity loss lives. The M-series word is narrower than its J-series counterpart, so a position forwarded from Link 11 to Link 16 may carry coarser resolution, fewer amplification fields, and a degraded track-quality indicator. The DLP has to make defensible default choices about what to drop and what to derive, and those choices are vendor-specific. Two DLPs from different suppliers, fed the same Link 11 input, will not necessarily produce identical Link 16 output.

6. the cost running estimate

The long pole in any Link 11 migration is terminal hardware, not software. Link 22 NIUs and modern MIDS terminals for Link 16 come from a handful of cleared suppliers with production measured in hundreds of units a year and lead times of 24–36 months. A program that needs new radios on twenty hulls is constrained by that pipeline far more than by any line of code.

Around the hardware sit three more cost centers that program managers routinely underestimate. Qualification: each national platform-radio combination needs security accreditation and interoperability certification, which is months of test events. Training: operators and maintainers have to be retrained from a serial-net mental model to a networked one. And the dual-running period — the months or years where the ship carries and powers both the legacy and the new link — costs space, weight, power, and crew time on equipment you are actively trying to retire.

7. translation fidelity

It is worth being precise about what translation actually drops, because "Link 11 to Link 22" sounds lossless and is not. M-to-F translation is the cleaner of the two: the F-series was designed to absorb M-series semantics, so most fields map directly, though F-series carries over-the-horizon routing metadata that has no M-series origin and is simply null on tracks forwarded up from Link 11. M-to-J is lossier: the J-series fixed-format words assume resolution and amplification the legacy M-series never carried, so forwarded tracks arrive with derived or default values in fields the originating link could not populate.

The distinction that matters operationally is weapons-quality versus track-quality coordination. A surveillance track surviving translation at track quality is fine for situational awareness. A weapons-quality engagement coordination — where the firing decision depends on the precision and pedigree of the track — cannot tolerate silent degradation across a link boundary. Every migration architecture has to identify which exchanges are weapons-quality and guarantee they either stay on a single link end to end or are translated by a path whose fidelity has been analyzed and certified.

8. a phased migration plan

The pattern that works is deliberate parallel running, never a flag-day cutover. Phase one installs the new link (Link 22 NIU or Link 16/JREAP) alongside the live Link 11 net and stands up the DLP to forward between them — at this stage the new link carries no operational weight and exists to be validated against ground truth the legacy link already provides. Phase two shifts primary traffic to the new link for selected track categories while Link 11 remains hot as a fallback, so any forwarding or fidelity defect surfaces with the old link still available to cross-check.

Phase three sunsets Link 11 against explicit criteria, not a calendar date: the new link must demonstrate equal or better track continuity, weapons-quality exchanges must be certified on the new path, and every coalition partner that needs the data must be reachable without the legacy net. Only when those gates are met does Link 11 come off the hull. The architectural enabler underneath all of it is the same one we recommend across every data link program — a dual-stack design with a canonical internal track model and versioned, hot-swappable protocol adapters for M-series, F-series, and J-series, so that adding or retiring a link is an adapter change, not surgery on the combat system. Build that boundary first, and the migration becomes a sequence of low-risk steps instead of a single irreversible jump.