Link 16 network design: NPGs, time slots, and the OPTASK LINK

A Link 16 network does not exist until someone designs it. The waveform, the terminals, and the message catalog are all standardized, but the allocation of time on the air is a finite resource that has to be apportioned deliberately, unit by unit and message by message, before the first platform enters the net. Get the design right and a hundred participants share a coherent air picture with sub-second freshness. Get it wrong and the net silently drops tracks, starves voice, or refuses net entry to half the force. This is an engineering walk through how a Link 16 network is actually built: the TDMA structure underneath it, the logical channels that ride on top, the math that decides who gets how much, and the operational message that ties it all together. It assumes you already know what Link 16 tactical data links are and want to know how to plan one.

1. the TDMA backbone

Everything in Link 16 timing nests inside one number: the epoch, 12.8 minutes long. Each epoch divides into 64 frames of 12 seconds each. Each frame divides into 1,536 time slots of 7.8125 milliseconds. That slot is the atomic unit of the network — a single transmission opportunity, long enough to carry one fixed-format message with its sync, packing, and propagation guard. Multiply it out: 1,536 slots per frame times 64 frames is 98,304 slots per epoch, and the whole schedule repeats every 12.8 minutes.

The 1,536 slots in a frame are not addressed as a flat list. They are interleaved into three slot blocks — A, B, and C — each contributing 512 slots, dealt round-robin so that slot index 0 belongs to block A, 1 to B, 2 to C, 3 to A again, and so on. This interleave is deliberate: it spreads any one participant's assignments evenly across the 12-second frame instead of clumping them, which keeps update latency smooth and avoids long silent gaps. When you allocate capacity you are really allocating slots within a block, and the block letter is part of the address.

2. Network Participation Groups

Raw time slots are not how operators think about a network. They think in Network Participation Groups — NPGs — which are logical channels layered over the slot pool, each carrying a defined class of traffic. A unit subscribes to the NPGs relevant to its role and ignores the rest. The standard NPGs are stable across the alliance: NPG 5 and 6 carry PPLI (Precise Participant Location and Identification, the J2 family); NPG 7 carries surveillance (J3 tracks); NPG 8 and 9 carry mission management and weapons coordination; voice rides on dedicated NPGs (typically 12 and 13) at 2.4 or 16 kbps.

Designing the network is, in large part, deciding which units participate in which NPGs and how much of each NPG's slot allocation each unit receives. A fighter needs PPLI and surveillance and one mission-management channel; it has no business transmitting on the air-control NPG that belongs to the AWACS. A ground air-defense fire unit may transmit on surveillance and weapons coordination but consume voice only as a receiver. The NPG is the unit of policy; the slot is the unit of resource. Good design keeps those two layers cleanly separated.

3. time-slot allocation

Slots within an NPG are handed out under one of two access modes. Dedicated access assigns specific slots to a specific terminal — slot A-7-2 belongs to ship X and to no one else, every epoch, guaranteed. It is collision-free and latency-deterministic, which is why surveillance and weapons traffic use it. Contention access lets a pool of terminals transmit opportunistically into a shared block, accepting the risk of collision in exchange for not pre-allocating per-unit capacity. PPLI from a large, fluid set of participants often uses contention so the designer does not have to name every transmitter in advance.

Slots are written in Time Slot Block notation: a triplet of block letter, starting slot index, and recurrence rate exponent, as in A-7-2. That reads as block A, starting at slot 7, recurring at 2 to the power of the third number per epoch — the recurrence number is an exponent, so the count climbs in powers of two: rate 0 is one slot per epoch, rate 6 is 64, rate 12 is 4,096. This compact triplet is how every assignment in the design is recorded.

When one frame's worth of slots cannot hold the required traffic, the design stacks nets: multiple logical nets share the same NPG by separating on the frequency-hopping pattern and net number, so several conversations coexist on the same time slots without hearing each other. Multinet operation lets a region run distinct air-control nets in parallel, at the cost of terminals that can only listen to one net at a time having to choose.

4. net entry and synchronization

A terminal cannot transmit until it knows what time it is, to within a fraction of a 7.8125 ms slot. Synchronization happens in two stages. Coarse sync aligns the terminal to the network's slot boundaries by listening to entry messages and establishing which slot is which. Fine sync then drives timing accuracy down to the sub-microsecond level needed for the frequency-hopping waveform to actually decode, by measuring round-trip time (RTT) against a reference.

That reference is the Net Time Reference — the NTR — one designated terminal whose clock defines network time. Every other participant synchronizes to the NTR, directly or through relay. Initial entry is the moment a new terminal listens for the NTR's transmissions, achieves coarse sync, exchanges RTT interrogations to pin down propagation delay, reaches fine sync, and only then begins transmitting in its assigned slots. Designate the NTR poorly — put it on a platform that leaves the operating area — and the whole net loses its time base. The NTR assignment is a first-order design decision, not an afterthought.

5. relay and range extension

Link 16 is line-of-sight UHF. Two surface combatants over the horizon from each other simply cannot hear one another, no matter how the slots are allocated. Relay bridges that gap. In active relay, a designated terminal receives a message in one slot and re-transmits it in a different, separately allocated slot so distant participants can hear it. In passive relay, the relaying platform's normal transmissions are themselves used by others as a timing and data reference without a dedicated repeat.

Active relay is expensive because it consumes slots — every relayed message needs its own transmission opportunity, so relaying an NPG can double that NPG's slot demand for each hop. A two-hop relay chain across a dispersed task group can quietly absorb a quarter of the available frame. The line-of-sight horizon — roughly 300 nautical miles between aircraft at altitude, far less for ships — is what forces relay in the first place, and the slot budget for it has to be reserved during design, not improvised when the net goes quiet at range.

6. capacity planning

Capacity planning is the arithmetic that makes or breaks a design. Start from slot loading: the percentage of the 98,304 slots per epoch already committed to dedicated assignments, relay, and voice. Then work the update rates. PPLI for a high-dynamics fighter wants a position report several times per second; for a slow-moving ground unit, once every few seconds is fine. Multiply each participant's required report rate by the slot cost and sum across the force, NPG by NPG.

Track capacity falls straight out of this. The surveillance NPG can carry only as many J3 updates per second as its allocated slots permit; divide that by the per-track refresh rate the mission demands and you have the maximum number of tracks the net can hold fresh. The fundamental tension is density versus latency: pack more participants and more tracks into the frame and either you raise slot loading toward saturation or you stretch each track's refresh interval. There is no free capacity — every slot given to one participant is a slot another cannot have, and the only honest answer to "can the net hold 600 tracks" is to do the slot math.

7. the OPTASK LINK

All of this design crystallizes in one document: the OPTASK LINK, the operational tasking message that directs how the data link network will be configured and operated for a given operation or period. It specifies the NTR, the NPG structure, the slot allocations per participant, the relay scheme, the crypto and net numbers, voice net assignments, and the initialization data every terminal must load before entry. It is the authoritative source of truth — if a terminal's initialization does not match the OPTASK LINK, that terminal will not interoperate correctly, full stop.

Producing the OPTASK LINK is the output of a planning workflow. The Network Design Load (NDL) is the machine-readable allocation that planning tools generate and that terminals consume. TDL planning products — network design tools used by the alliance's data link management cells — take the force list, the operating geometry, and the mission's traffic requirements and compute a consistent slot allocation that fits inside one epoch. Those tools exist because doing the slot arithmetic by hand for a hundred participants across a dozen NPGs is error-prone in exactly the ways that strand units off the net. Many of the messages the OPTASK LINK provisions for are catalogued in the J-series message catalog, and the design has to reserve slot capacity for each one the operation needs.

8. common design mistakes

The same handful of failures recur across programs. Over-subscribed NPGs are the most common: the designer sums the desired update rates, finds they exceed the slots allocated to the NPG, and rather than re-budget simply hopes contention will sort it out — so PPLI collides, position reports drop, and the picture degrades exactly when participant density is highest. Ignored relay budget is next: the design assumes full connectivity, the force disperses, and the relay slots that should have been reserved up front were never allocated, so over-horizon units fall silently off the net.

The third is mismatched initialization data: one terminal loaded with a crypto variable, net number, or NPG assignment that differs from the OPTASK LINK. The symptom is maddening — a unit appears healthy, achieves sync, and still cannot exchange tracks with the rest of the force, because it is effectively transmitting into a different logical net. Disciplined configuration management against a single authoritative OPTASK LINK is the only cure. The lesson across all three is the same: Link 16 capacity is finite and the math is unforgiving, so the design has to be done deliberately, validated against the slot budget, and held in sync with the tasking message that the whole force loads from.