Join War on the Rocks and gain access to content trusted by policymakers, military leaders, and strategic thinkers worldwide.
“[T]he exigencies of desperate survival do stimulate wise leaders to seek such unorthodox solutions.”
—Barton Whaley, Stratagem
On the night of Jan. 5, 2025, my first mission in the Pokrovsk command post was to make a fake radio call. We knew Russian forces were listening in. We did not know whether the position our men were due to occupy the next day, the Homar trench at Kotlyne, was still in our hands or if the enemy had already taken it. I did as instructed: I pretended to be in the trench and called the command center for help.
Two minutes later, through the thermal camera of the drone hovering above the position, we saw two white silhouettes emerge from a bunker, along with a third, unarmed person. The three of them began patrolling and checking the trench. The Russians intercepted our call and sent the two soldiers holding the position, apparently with a prisoner, to verify. We had induced the adversary to alter its behavior, and we had read the position’s true status in its reaction.
The fighting at Pokrovsk hit us with brutal clarity as an instance of mass autonomous warfare — a battlefield dominated by advanced tech and unmanned aerial assets, reshaping traditional tactical doctrine. In the fighting that followed, we gained further advantages the same way, pretending to hold portions of terrain while we were evacuating the survivors elsewhere.
Three features of that episode generalize. First, the fake call served as reconnaissance. Second, it produced a kinetic response and exposed the enemy’s situation (they were already infiltrated, listening to our communications, and holding a prisoner). Third, it concealed our real objective by diverting the enemy’s attention to the wrong sectors. These lies created a vital informational fog. At platoon scale, deception costs little; forces can improvise it from equipment already in hand, and it returns information as well as protection. This is an issue I care deeply about due to my wartime experience, which led me to start a company called Fog Defence, which develops decoys: firmwares and state machines that replicate tactical signatures.
On a battlefield made transparent by sensors, concealment alone no longer protects maneuver. One solution is deception: producing a false narrative (decoys, fake maneuvers, misleading intelligence), in coordination with real movement, so the adversary is induced to make wrong assumptions or can no longer tell what matters from what does not. Today’s military deception target is not only the enemy commander but the entire automated chain of sensing, classification, and decision, and the tempo at which that chain produces decisions. An up-to-date doctrine should address this sensor-fusion chain as well as the adversary decision-maker, with explicit concern for economic and operational profitability. I define profitability here as two measurable things: whether the adversary spends more than the deceiver to engage a decoy, and whether the deception buys freedom of action. At Kotlyne I observed a human reaction to a fake call. The effect on automated classifiers is my proposal, not an observed result — and not every targeting chain is automated.
Analysts now widely recognize the problem of surface maneuver and battlefield air interdiction. If one can no longer hide while moving, one should muddy the waters and push the adversary toward false assumptions. Diversion has never been so necessary in the face of drone interdiction on open terrain.
Turner Shaw’s recent article, “Train to Lie: Small-Unit Deception on a Transparent Battlefield,” correctly identifies a doctrinal gap at squad, platoon, and company levels. I share his assessment that concealment alone is insufficient under persistent surveillance across multiple environments. Shaw focuses on institutionalizing small-unit deception. I propose nesting deception across echelons — from section to corps — into a single coherent system, and emphasize that stratagems address the automated classification chain rather than the human commander alone.
The attritional logic of deception is easiest to see in the air. Since 2024, Russian strike packages have relied on unarmed decoys: Gerbera foam drones and wooden Parodiya drones, both fitted with a Luneburg lens that gives a radar signature close enough to an armed Shahed or Geran to be treated as a threat. The media has reported that these decoys made up 50 percent of swarms launched at Ukrainian territory, with the rest armed. The ever-expanding Alabuga complex manufactures all of it, keeping those fake drones’ price below $10,000. A Western surface-to-air interceptor costs several hundred thousand to several million dollars. That lopsided exchange ratio explains why Ukraine developed cheaper interceptors such as the Wild Hornet’s Sting drone ($2,500) and manufactured its own long-range decoy, OM Defence System’s Spectr, to exhaust Russian air defenses as well.
This documentation exposes that decoys can drain interceptors. A mimetic signature may exploit a window of first exposure, but a system trained on confirmed examples will label a known fake signature. Neither outcome is automatic. A decoy will produce either a false picture if it succeeds, or, in the worst case, it will consume analyst attention or classifier bandwidth before rejection. Both are hypotheses, not demonstrated battlefield behavior, as the decisions of classifiers are not publicly observable. The length of the window is unknown but likely short, as adversaries retrain and adapt.
The arithmetic is tighter on the ground, because the decoy should be cheaper than the low-cost munition that engages it. High-end inflatable decoys of armored vehicles can range from $10,000 to $100,000, and their return on investment lies in the preservation of real assets — against, for example, the cost of a Javelin round, which itself can vary depending on who the buyer is and when they buy it (as of Fiscal Year 2021, the Department of Defense’s unit cost for a G-model Javelin missile was $216,717). But that logic does not carry over to a drone-saturated battlespace. A first-person view drone with a thermal camera cost roughly $2,500 in 2024. According to Robert “Magyar” Brovdi, commander of the Unmanned Systems Forces, mass production of first-person view kamikaze drones now puts them at $400 per unit, excluding a specific sensor and explosive payload.
The ideal cost target for an attritional decoy aimed at consuming first-person view drones would be a fraction of that $400 figure: a radio-frequency emitter built from commercial off-the-shelf components using a few metal-oxide-semiconductor field-effect transistors to imitate a thermal signature would need to meet that target. No documentation shows such low-cost disposable emitters, matching the price tag of first-person view drone builds, in actual use. But the need for electromagnetically mimetic lures has emerged since the Russo-Ukrainian War demonstrated how signature management has become critical. Beyond attrition, mass mimicry would buy two things: operational ubiquity, with real assets mixed among false signatures, and “corridors” for real offensive assets, when defenses are saturated. Aerial warfare shows both. On the ground, the mechanism is the same, but has to be demonstrated across the full spectrum — beyond the use of a single decoy signature in the physical or aural spectra, like dummies or loudspeakers dropped by drones. The addition of transistors and radio-frequency emitters to inflatable armor enriches the decoy’s signature in order to lure machines, but as we discussed, these deception tools are not low cost.
Even if a ground decoy exceeds the cost of the drone that investigates it, the deception can still be profitable if it produces a tactical effect the defender values more highly than the outlay: opening a corridor, reinforcing an existing enemy belief (Magruder’s principle states that it is easier to reinforce an existing belief than to create a new one), generating surprise, or saturating cognitive and classification chains. The return then reflects restored freedom of action, not solely expended munitions.
Saturation consumes resources, but it does not produce a specific false decision. A more consequential form of deception builds a coherent fabricated picture that leads the adversary to reason in a chosen direction. This is reflexive control: not simply hiding information or creating noise, but shaping the informational conditions from which the adversary constructs its own decisions. Reflexive control emerges from the broader Soviet and Russian practice. In essence, it means feeding the adversary tailored information so that it reaches, by its own reasoning, the deceiver’s intended decision.
Operation Vivaldi illustrates this logic. Conducted in September 2026 and orchestrated by Ukraine’s Third Army Corps commander Andriy Biletsky (who declared that “deception had been a key factor of the success”), the first sequences aimed at blinding the adversary. Mid-range strikes degraded Russian electronic warfare systems and targeted the adversary’s drone-piloting teams. According to deputy commander Maksym Zhorin, strict information silence and a “protective dome” of electronic warfare systems (direction finders and jammers) helped to disrupt and suppress around 20,000 Russian drones. Ukrainian forces then feinted toward deploying “narrative-coherent” intoxication and ultimately retook the Nove settlement and its heights. A special unit within the corps, dedicated to ‘’operational security,” undertook the creation of “a picture on the battlefield that does not correspond to the reality,” shaping the adversary’s representation through fake signatures and operational narratives (Third Army Corps’ report of September 25th). The result is not only territory liberated, but maneuver restored through deception and tactical consistency.
Cross-echelon stratagems, including those that unit-level practitioners use, work only when forces genuinely share situational awareness and strategy within an electromagnetic order of battle. The Delta system, Ukraine’s real-time mapping system, aims to collect inputs from multiple battlefield sensor feeds to build a shared operational picture. Delta enables comprehensive mapping of the battlespace and helps execute situationally aware maneuvers and stratagems.
Joint Publication 3-13.4 defines military deception as actions to mislead adversary decision-makers, “thereby causing the adversary to take specific actions (or inactions),” and counts dummy equipment and deliberate emissions among its means. Field Manual 3-13.4 adds the category of tactical deception. Behind both stands Cold War scholarship: Barton Whaley’s typology of hiding (masking, repackaging, dazzling) and showing (mimicking, inventing, decoying), and Donald C. Daniel and Katherine L. Herbig’s A-type (ambiguity-increasing) against M-type (misleading) distinction in “Propositions on Military Deception.” In “Military Deception in Peace and War,” collected in War, Strategy, and Intelligence, Michael Handel argues that deception is a rational force multiplier even when it misses its original goal.
The existing framework still strains against a machine-read battlefield. First, the deceiver augments the target. The joint publication’s principles aim the deceiver at the decision-maker’s actions rather than his beliefs. That presumes a human whose picture the deceiver can model. On a transparent battlefield, the first reader of a signature is a classifier, and the human sees only what it lets through. The decision-maker is a chain consisting of a sensor-fusion engine, a classifier, and an operator, and a decoy can saturate or fool each link on its own terms.
Second, the ranking of methods has also evolved. The deceiver addresses an intermediate target: the machine that constructs the operational picture, as in Biletsky’s operation. This suggests an ambiguity-to-misleading-type deception continuum in which ambiguity-type effects first increase within the sensing and fusion chain (false signatures, competing tracks) in order to saturate it. If the deceiver then resolves the resulting uncertainty in favor of a deliberately induced false interpretation, the effect moves toward misleading-type. The continuum runs from disrupting information processing to shaping the false picture that emerges. The deceiver induces the machine to produce an erroneous representation of the battlespace. Deception moves from hiding a signature to manipulating the detection-fusion-classification-decision cycle. On its commander’s account, Vivaldi aimed at the misleading-type endpoint through cumulative ambiguity-type stratagems.
Third, deception has decentralized. Doctrine requires central planning for military deception so that every means portrays the same story. A section’s emitters that contradict the brigade’s story expose both, and the more channels a story runs through, the harder it is to sustain. Shaw’s remedy is preapproved small-unit deception. It settles who may deceive, but not how local signatures stay consistent with the higher story and with one another. That coherence problem demands signature management nesting: one story, areas of responsibility that flow downward, and a shared picture where every signature aligns.
Finally, the tempo of deception has also shifted. A classifier shortens intervals to seconds and gives the deceiver something back: The reaction becomes visible almost at once, as at Kotlyne, when two Russian soldiers emerged two minutes after the call and the thermal camera hovering above could see them. At this tempo, forces no longer execute a deception plan against an expected response — it becomes a loop they sense, test, and iterate. When sensing and response approach machine speed, deception should become a continuously updated interaction. Otherwise, the advantage of the cycle belongs entirely to the faster side.
The most serious objection is that sensor-fusion chains and adaptive classifiers will eventually recognize repeated patterns. A decoy that works today can become an indicator of deception tomorrow. The issue is therefore one of adaptation rate. A digitally reprogrammable emitter can alter its signal parameters extremely rapidly, while a defender relying on data collection, labeling, retraining, and redeployment may require longer to adapt. But model learning and cross-corroboration can shorten the defender’s response time. A useful comparison is between the deceiver’s variation period and the defender’s effective adaptation cycle — the advantage persists only while the former is shorter than the latter. The deceiver should also maintain coherence across sensors and continuous variation to keep the advantage. Eventually, the adversary may not need to recognize the deception at all — he can probe cheaply or simply stop reacting as the deceiver intended. The same measure-countermeasure adaptation cycle seen in drone innovation governs the contest between luring emitters and the classifiers they aim to deceive.
A doctrine built on deliberate signatures leaves no room for improvisation. Decoys provoke a response that can endanger friendly forces nearby and escalate uncontrollably. Each deployment requires a documented perimeter: what the signature is, why it exists, how far it extends, for how long, and what response the deceiver expects it to draw. The law of armed conflict permits ruses (decoys, feints, military disinformation) and prohibits perfidy, which invites the adversary’s confidence in the deceiver’s protected status, only to betray that confidence by killing, wounding, or seizing the adversary. Article 37(2) of Additional Protocol I settles only that question — a ruse is lawful if it is not perfidious and breaches no other applicable rule — but it does not make any particular deployment lawful. Distinction, proportionality, and precautions against the effects of attacks apply independently, and deception that deliberately attracts fire carries legal exposure wherever civilians may be present. A Western doctrine that institutionalizes signature deception should therefore specify: a ban on imitating protected signals (medical, humanitarian, civil aviation, and maritime distress) and on mimicking civilian infrastructure; a documented civilian-presence assessment, with abort conditions; exclusion zones sized to the response provoked; power and spectrum limits; time limits with self-expiry; coordination with adjacent friendly units; an accountable authorizing commander; and a legal review of the platform and its employment. Without these safeguards, the risk of lethal and legal exposure is real.
A force that saturates its area with false thermal, radio-frequency, and acoustic signatures also degrades the picture available to its own units and coalition partners. One nation’s decoy fleet is another nation’s false target set. Commanders should therefore deconflict stratagems inside the shared situational awareness that systems such as Delta aim to provide. Units should log, geolocate, time-bound, and share every deployed signature with their partners. Otherwise, the risk of operational fratricide and coalition friction is high.
Individual signature management can shift to the combat-group level, so that every small team treats its own thermal, radio-frequency, acoustic, and behavioral footprint as a managed variable. At the company level, units should orchestrate limited fleets of signatures (radio-frequency emitters, false thermal, acoustical, and physical decoys) in coordination with real movement. At the battalion and brigade levels, the focus shifts toward reflexive control stratagems. Offensive and defensive electronic warfare capabilities should routinely accompany ground platforms. Acquisition authorities should issue targeted calls for proximity camouflage systems and low-cost active decoying solutions fieldable at scale. Planners should prioritize the survivability of the asset, whether it is aerial, terrestrial, or naval, through “operational ubiquity”: a mimetic signature that replicates its presence on the adversary’s battlespace picture.
And none of this is free. A company carrying emitters and decoys is carrying less of something else. To illustrate the problem, let me relate an anecdote. Before we moved to the defensive positions near Pokrovsk, each soldier chose what to pack. Ammunition was unlimited, and we each carried our five fragmentation grenades. We could also take a portable anti-drone system with a spare battery. It easily weighed five kilograms, and with its antenna and battery it filled two-thirds of a 50-liter “24-hour bag.” Some of us left this kit behind and took water and ammunition instead. Whatever gadget is invented today to spoof an adversary by imitating the signature of a soldier or a vehicle, carrying it should be neither burdensome nor disproportionate. Its cost to institutional buyers should also fit a favorable equation: proportionate to what it protects or enables, or to its attritional value, meaning the cost of the munition that may eventually engage it.
Western armies have cut deception and electronic warfare specialists to very little over three decades. Reintroducing the capability requires either internal reallocation or new capacity. The institutional question remains open: Should signature fleets be organic to maneuver units, or should a higher echelon centralize and task-organize them downward? A credible proposal should state approximate costs, identify what it displaces, and name the institutional owner. Ownership could sit at battalion level, within decentralized electromagnetic spectrum operations cells. Such a signature management cell can hold the library and reprogram the emitters while companies employ them. This follows from the rate argument above: Variation needs central control to be fast and coherent, while employment requires decentralization to stay credible and local. Organic company fleets would give speed of use, but neither coherence nor renewal, which is a specialized responsibility.
Tunnels and deeper fortification buy time but do not restore freedom of action on the surface. Extreme lethality on the surface need not become fate. The same logic already applied to long-range effectors extends to the ground, too: multiplying signatures so that the real asset’s presence is no longer the only thing the sensor sees. The future of military deception rests on addressing a sensor-fusion chain with concern for economic and operational profitability. The night at Kotlyne was a miniature of this principle: one false call, one reaction read through a thermal camera, and later a handful of survivors who got out because the enemy was looking elsewhere.
The operational impasse created by the air exclusion zone effectively narrowed our options to the point where practitioners had no choice but to innovate in the field of deception and tactical stratagems. Just when we believed we faced certain doom, as in Pokrovsk in January 2025, the prospect of achieving tactical victories through deception emerged as the last viable course of action. Beyond and throughout the electromagnetic spectrum battlespace, the ultimate battleground remains the adversary’s cognitive system, and deception exists to manipulate it.
Édouard R. Maurin, call sign FIDES, is a French writer and veteran who served with an international volunteer section attached to a Ukrainian special-purpose battalion near Pokrovsk in 2024 and 2025. His war journal, La Chair à Drone, recounts that deployment. He is developing commercial off-the-shelf tactical deception hardware and software.
Image: ArmyInform via Wikimedia Commons
.webp){kind=link}