The Network Is the Weapon

 An airborne warning and control system detects a target hundreds of kilometers away—well beyond the radar range of any fighter in the formation—and tracks (the calculated path or trajectory of a moving object generated by connecting consecutive radar detections in a geodetic frame of reference) are transmitted via encrypted datalinks to a fighter, which launches a missile it could never have fired on its own capabilities. The missile flies the first phase on inertial navigation, receiving mid-course updates from the launching aircraft or from other networked platforms. Only in the terminal phase does the missile’s own seeker activate. In this scenario, the aircraft that detected the target was not the one that killed it. The platform that provided guidance was not the one that launched the weapon. Sensors and effectors are effectively decoupled, in a principle called “any sensor, any shooter”.

But what does connect all these systems together?

The Datalink Ecosystem

Western tactical datalink architecture has grown over decades, each generation adding its own link optimized for the requirements of its time. The result is a complex landscape, and quite far from perfect.

Link 16 is the (legacy) baseline. A frequency-hopping UHF waveform dancing across 51 frequencies in the 960–1215 MHz band, carrying standardized messages (J-messages) using Time Division Multiple Access (TDMA). The physical hardware is the Multifunctional Information Distribution System (MIDS), a family of terminals fitted to everything from F-16s to F-35s to Patriot batteries to frigates. Link 16’s value lies in its proven interoperability. An Australian F/A-18F, a British frigate, a Norwegian ground radar, and a French AWACS can all see the same tactical picture because Link 16 standardizes what a track looks like, how identification is shared, and how engagement status is communicated. Needless to say, such links need to be secure. Link 16 uses two independent layers of cryptographic protection. Message Security (MSEC) encrypts the message contentat the link layer. Transmission Security (TSEC) encrypts at the physical layer, controlling the waveform itself; that is, the pseudo-random frequency-hopping pattern, chip scrambling sequence, symbol interleaving, and message jittering are all driven by the TSEC crypto variable. Both use Traffic Encryption Keys (TEKs) that rotate on a daily cryptographic period, loaded into the terminal's Secure Data Unit (SDU). On modern platforms like the F-35 and F-22, this is handled by embedded programmable COMSEC modules. Key distributionis managed through the Electronic Key Management System (EKMS) or its successor Key Management Infrastructure (KMI), with Key Encryption Keys (KEKs) protecting the TEKs in transit. The entire key management architecture is governed by CJCSM6520.01B, the Link 16 Joint Key Management Plan.

However, Link 16 has several limitations. The data is surveillance-grade only, which is sufficient for situational awareness, but too imprecise and too slow-updating for direct weapon guidance. Also, Link 16 waveform is omnidirectional, meaning it can be detected and located by an adversary. Its jam resistance, while substantial, is still finite. An adversary with high-power broadband barrage jamming across the full L-band can degrade it, though the cost in jammer power would be significant.

Link 22, developed by a seven-nation NATO consortium (US, UK, France,Germany, Italy, Canada, Spain), extends the tactical networking baseline Beyond Line Of Sight (BLOS). Operating on both High Frequency (HF) and UHF waveforms, it enables units up to 1,000 nautical miles (~1800 km) apart to participate in the same tactical picture via HF propagation, which is critical for naval operations where ships may be spread across large bodies of water. Its dynamic routing and automated network management make it more resilient than Link 16’s fixed TDMA structure,and it was designed from the outset to complement and interoperate with Link 16.

When it comes to actual fire control, systems like Cooperative Engagement Capability bring the determinism for engaging targets. Conceived by the Johns Hopkins University Applied Physics Laboratory (APL) in the early 1970s and entering fleet service in the late 1990s, CEC distributes raw, unfiltered radar measurement data rather than the processed tracks that Link 16 carries. Every CEC-equipped platform runs identical fusion algorithms on the combined measurements, producing a composite track picture that is richer than what an individual sensor’s output can give, and precise enough to guide effectors. CEC operates in C-band (4–8 GHz) via the Data Distribution System (DDS), a jam-resistant link using a phased array antenna that communicates point-to-point between cooperating units through rapid, directional beam-switching. Data rates are estimated at two to three orders of magnitude above Link 16's ~25–100 kbps. Each node runs identical fusion algorithms in a Cooperative Engagement Processor (CEP), producing a composite track picture precise enough for fire control, allowing a ship to launch a missile at a target it has never seen on its own radar. CEP is line-of-sight only; thus, airborne platforms like the E-2D Hawkeye can extend the network's reach by serving as both elevated sensor and relay node. It's built by Raytheon, and designated AN/USG-2 (shipboard), with variants. As a key pillar of the Naval Integrated Fire Control-CounterAir (NIFC-CA) capability, CEC allows integrated fire control across the fleet, transforming a carrier strike group from a collection of ships into a distributed weapon system where every sensor contributes and every weapon canbe assigned to any target regardless of which platform carries it. Is space anywhere in the fire control picture? Although space plays a part in networked warfare, the challenge of tracking low-altitude maneuvering targets from orbit, with the update rates and precision that CEC provides, remains somewhat unsolved. This would require persistent EOIR/radar monitoring from LEO with real-time downlink to multiple vessels and vehicles with compatible waveforms, security, and stealth. Let's address the latter next.

The stealth layer

Fifth-generation stealth aircraft present an interesting contrast with Link 16: the whole point of stealth is to minimize electromagnetic emissions, and Link 16 is, as we said, an omnidirectional broadcast. The solution is a different class of datalink.

The F-35’s Multifunction Advanced Data Link (MADL) uses narrow, directional beams from phased array antenna assemblies embedded in the airframe skin. Each transmission is tightly focused on a specific receiver, making it extremely difficult to intercept or detect. MADL provides significantly more bandwidth than Link 16, enabling F-35s to share a richer sensor-fused picture covertly, including radar tracks, infrared (IR) detections, and threat classifications.

The F-22’s Intra-Flight Data Link (IFDL) serves the same purpose within Raptor formations. Interestingly, they are completely incompatible with each other. MADL and IFDL use different waveforms, different frequencies, different protocols. The F-22’s planned MADL upgrade was cancelled in 2011 due to cost and technology maturity concerns. The incompatibility arose from thetwo systems being two decades apart, with different requirements and use cases. The F-22 can receive Link 16 but cannot transmit on it (to preserve stealth). The F-35 can both transmit and receive, but transmitting also breaks stealth. It is still interesting that the two most advanced fighters in the Western inventory have no bidirectional communication path between them without the hel pof an external actor, which typically consists of using gateway platforms that fly above them and carry multiple link terminals and translate between them. In 2021, Lockheed Martin Skunk Works demonstrated Project Hydra: a U-2S at 70,000 feet (~21 km) carrying an Open Systems Gateway (OSG) that connected an F-22 and five F-35s via simultaneous IFDL and MADL reception, making the first bidirectional fifth-generation communication between the two types. The E-11A Battlefield Airborne Communications Node (BACN) performs a similar translation role for Link 16 and other tactical links on a Bombardier business jet. Needless to say, gateways are vulnerable and a single point of failure. Additionally, the translation adds latency.

The Orchestra and Its Conductor

Datalinks carry the data. However, in a complex operation—for instance, a Suppression of Enemy AirDefenses (SEAD) campaign or a multi-nation strike—someone must orchestrate who transmits what, when, and to whom. That role belongs to AWACS.
The air battle managers aboard the E-3 Sentry (or its successors, the E-7 Wedgetail and SaabGlobalEye) build and maintain the air picture across a 400+ km radius, tracking every aircraft—friendly and hostile—simultaneously. They deconflict the electromagnetic operations of multiple platforms: ensuring that one jammer doesn’t blind another platform’s targeting sensor, that electronic warfare (EW) assets complement rather than interfere with each other, that the collective electromagnetic footprint serves the mission.
In a SEAD mission, an RC-135 Rivet Joint collects Signals Intelligence (SIGINT), the EA-37B Compass Call jams enemy command networks, EA-18G Growlers provide close-in radar jamming, and F-16CM Wild Weasels hunt active surface-to-air missile (SAM) radars with AGM-88 High-Speed Anti-RadiationMissiles (HARMs), while AWACS coordinates the entire sequence. It provides threat warning while the EW platforms focus on their specific tasks. It dynamically retasks assets as the enemy reacts. It is the conductor of an electronic warfare orchestra that would produce just random noise without coordination.

Without AWACS, each platform operates as an individual, unable to synchronize into a coherent campaign. This is why the current US AWACS fleet crisis—16 ageing E-3 Sentries with a 56% mission-capable rate— represents a systemic risk to the entire Western EW in general and its SEAD capabilities.

The Fleet as a Distributed Weapon System

On the naval side, a carrier strike group connected through CEC operates as a single distributed weapon system. An incoming anti-ship cruise missile is detected by an E-2D Advanced Hawkeye at range. CEC distributes the track to every Aegis-equipped ship. The Aegis combat system automatically assigns the engagement to the ship with the best geometry, which is not necessarily the closest one, but the one with the optimal firing solution given missile kinematics and defensive coverage. The SM-6 launches, guided by CEC data from the E-2D. If it misses, a second ship engages. The entire sequence is coordinated automatically across the fleet.

This level of automation and integration—where the network itself makes engagement decisions faster than any human could—is what the Air Force and Army are still working toward through Joint All-Domain Command and Control (JADC2) and IBCS. The Navy went for it because the anti-ship cruise missile threat demanded reaction times measured in seconds, not minutes, where human-in-the-loop coordination was simply too slow to be considered.

The SM-6 is the weapon that makes the naval network lethal across missions. Carrying an active radar seeker derived from the AIM-120 Advanced Medium-Range Air-to-AirMissile (AMRAAM), it can engage targets the launching ship cannot see, flying to a CEC-derived position and acquiring autonomously. It performs Anti-AirWarfare (AAW), Anti-Surface Warfare (ASuW), and terminal Ballistic Missile Defence (BMD), selectable by software and targeting data rather than hardware.

The advantages of naval platforms in the networked warfare role are considerable when it comes to on-board power, a topic we discussed already. An Arleigh Burke-class destroyer generates roughly 7,500 kW of electrical power, which is approximately fifty times the F-35’s output. Ships stay on station for weeks, providing persistent network presence that airborne platforms can only achieve through rotation. And the fleet carries surface search radars, air search radars (AN/SPY-1 or the newer Active Electronically Scanned Array AN/SPY-6), sonar arrays, helicopter-borne sensors, EW intercept systems, all feeding the same CEC/Link16/combat system architecture.

Convergence: IP, Space, and What is Next

Two forces appear to be reshaping the tactical datalink ecosystem: the migration to Internet Protocol (IP)-based networking, and the proliferation of Low Earth Orbit (LEO) satellite constellations as communications backbone.

The Space Development Agency’s (SDA) Proliferated Warfighter Space Architecture (PWSA) is a mesh network of optically connected satellites in LEO. In late 2023, SDA demonstrated the first-ever Link 16 network entry through a space-to-ground connection.

Separately, Starshield (the military derivative of Starlink) was tapped for testing data connectivity to F-35s in flight, which would bring substantially higher bandwidth compared to traditional military SATCOM.

An IP-based mesh constellation in LEO would increasingly serve as the transport backbone, with legacy datalinks riding over it and line-of-sight constraints eliminated. Link 16 traffic relayed through space means an F-35 over the Arctic can participate in the tactical picture without a chain of airborne relay gateways. A destroyer in the Mediterranean can share data with a ground station in Germany via satellite rather than vulnerable surface relays.

However, IP convergence does not solve all problems. IP networking remains non-deterministic, which is acceptable for situational awareness but still problematic for fire-control. Also, any terminal transmitting to a satellite is a radio emitter, incompatible with stealth operations. Last but not least, a single-vendor dependency for the military’s communications backbone introduces risk current conflicts have already illustrated.

It all indicates the specialized links will remain, while the architecture above them will increasingly be software-defined and IP-converged. The most likely end game is an IP network where each platform translates its native datalink format into standardized IP datagrams at the edges, and the satellite routes packets, agnostic to the source. Where the translation happens on the aircraft, not in orbit. No cryptographic systems in space, no cross-protocol translation on a cheap satellite with a loosely controlled supply chain, no multi-level security challenge in LEO.

The AI gap

The datalink ecosystem discussed here was designed around the rationale that what's important operationally is processed data. An AWACS operator works on radar tracks, never on raw radar returns. All in all, the architecture optimizes for transmitting processed results, minimizinng bandwidth. The F-35 generates extraordinary sensor data internally, including SAR imagery, infrared panoramas, and more. However, what leaves the aircraft via MADL or Link 16 is strictly processed tactical information. The raw sensor data stays onboard until the aircraft lands and is physically downloaded. This means that all raw data generated is not exactly available for a potential MLOps. Training a model to classify mobile SAM launchers from radar imagery requires thousands of acquisitions from multiple geometries. Currently, the only system in the architecture designed for raw sensor streaming—the Common Data Link (CDL), operating at up to 274Mbps—is carried by dedicated Intelligence, Surveillance, and Reconnaissance (ISR) platforms (RQ-4 Global Hawk, U-2, E-8 JSTARS) but is not supported by the F-35, for stealth reasons. This would starve any ML data pipeline, or make it prohibitively slow. However, there appears to be emerging concepts that could eventually give the chace to transmit data with high bandwidth while keeping stealth capabilities. For instance, the F-35's AN/APG-81 AESA radar is, at its core, a phased array of solid-state transmit/receive modules that can be electronically steered. Therefore, in principle, those same modules could be repurposed as a directional data transmitter, pointing a focused beam of data at a receiver rather than radiating radar pulses. This is sometimes called AESA comms or Radar Common Data Link (RCDL). The concept would allow the F-35 to use its existing radar aperture as a high-bandwidth,directional, LPI data link, transmitting SAR imagery or other sensor products in a narrow beam to a designated receiver (a ground station, a ship, an otheraircraft) without adding any extra hardware. The practical challenges are that the radar can't simultaneously search for targets and transmit data (you're time-sharing the aperture between its radar mission and its comms mission), and the receiver needs to be equipped with a compatible terminal.

All in all, the network is the actual weapon. However, currently, not a future-proof weapon. AI-augmented warfare will require an order of magnitude more bandwidth while keeping the stealth capabilities intact. It all points to optical networks, a topic we barely scratched today, and that will be a topic of a future piece.

Photo: Bombardier E-11A 11-9001 at Dubai Airshow 2021 (Credit: user Mztourist, Wikimedia Commons) (CC 4.0)