The next air raid will be detected by an object that cannot see the sky.
A microphone fixed to a utility pole will hear an engine beneath the radar horizon. Kilometres away, a thermal camera will find a warm point moving against cold cloud. A radio receiver will notice an unfamiliar transmission. A cellular tower will illuminate the same aircraft accidentally, its ordinary signal returning with a faint new echo.
Each observation will be inadequate. Software will place them on the same map, compare their time and direction, and decide that they describe one machine. Only then will an active military radar wake, look into a narrow piece of sky and give an interceptor the precision it needs.
Ukraine has already built the first outline of this system while defending its cities from Russian drones. Iran has supplied the opposite lesson: an air-defense network concentrated in prominent radars and command nodes can be mapped, attacked from inside its own territory and dismantled at the beginning of a campaign. The two wars point towards the same architecture. Air defense will spread its senses across the country so widely that destroying any one of them changes almost nothing.
ParallaxSee forecasts that by the end of 2032, at least three countries beyond Ukraine will operate national or theater-wide air-defense sensor webs that fuse multiple passive sensing methods with military radar tracks, use automated software to construct a shared air picture, and directly cue weapons or electronic countermeasures towards verified threats. House confidence: 84%.
The radar dish will remain. It will become one alert, intermittent member of a much larger organism. The future air-defense system will possess something its predecessors never had: a nervous system spread across the land.
01 — Radar will become the sharp eye inside a much larger body.
A conventional radar performs an elegant act. It sends radio energy into the sky, measures the returning reflection and calculates an object's range, direction and speed. Large military radars can survey immense volumes of air and produce the precise tracks required for long-range weapons. That power secures their place in the future network.
Their physics also imposes a bargain. A radar transmits, so an adversary can search for its emission. Terrain and the curvature of the Earth hide low aircraft below its horizon. Small composite drones return little energy. One exquisite machine must process birds, weather, decoys, cruise missiles and cheap aircraft arriving together. If it is destroyed, jammed or forced to move, the defended region can lose a large part of its vision at once.
The new architecture gives the radar a more survivable job. Passive nodes search continuously without advertising themselves. They report a rough direction, a sound, a heat signature or a disturbance in somebody else's radio transmission. Software correlates the reports. The radar then illuminates a smaller volume for a shorter period, supplies engagement-quality measurements and changes position or falls silent again.
The Center for Strategic and International Studies calls this mesh sensing: a thicket of high- and low-end sensors in which proliferation, distribution and control of emissions make the whole air-defense system harder to blind. Its modelling finds that passive sensors combined with active radar can extend coverage, conserve radar resources and help distinguish false targets from real ones. CSIS on mesh sensing for air and missile defense
The sensor web enlarges radar's power. It tells the sharp eye where and when to look.
02 — Ukraine has given a country thousands of ears.
A Shahed-type attack drone is slow, low and loud. Those characteristics once looked primitive. They have become a detection opportunity.
Ukraine's Ministry of Defence says its Sky Fortress network now contains more than 20,000 acoustic nodes connected through mobile communications. A second system, Zvook, concentrates sensors along approach corridors near the Russian border where radar detection is most difficult. Machine-learning models compare the sound of an approaching engine and propeller with a changing library of signatures. When Russia modifies the noise, the models can be retrained. Ukrainian Ministry of Defence on its detection architecture
The ministry describes the doctrinal change, in translation, as a move from “see far—respond centrally” to “see everywhere—respond locally and through the network.”
This is larger than a clever microphone. Acoustic reports join radar, thermal cameras, visual observation posts and digital command systems. A rough warning reaches a mobile fire team or interceptor-drone unit positioned near the predicted route. A sensor too imprecise to guide a weapon can still supply the minutes and direction that make another sensor useful.
The scale of the pressure forced the architecture to mature. A NATO Ukraine Insights report says Russia launched 53,961 Shahed-type drones during 2025, including about 31,578 strike variants. It records acoustic, electro-optical and radar warning feeding automated command systems, while interceptor drones accounted for roughly one-fifth of destroyed one-way attack drones by the end of that year. NATO's April 2026 Ukraine Insights
NATO's public study of acoustic sensing reaches the same conclusion: extend the network, integrate it with radar and electronic warfare, strengthen it against cyberattack and use AI to filter noise and classify targets. NATO study of acoustic sensors in Ukraine
Ukraine did not wait for a perfect sensor. It multiplied imperfect ones until the gaps began to overlap.
03 — Iran demonstrated why the sensing system must survive the first night.
An air-defense radar is both a shield and a target. The opening of the June 2025 Israel-Iran war made that dual identity unmistakable.
Israel prepared its air campaign with intelligence, electronic warfare, satellite observation and small precision-strike systems placed inside Iran. CSIS's comparison of the Iranian and Ukrainian campaigns reports that internal drones and missiles struck Iranian air-defense assets from inside their defensive envelopes. In the following 24 hours, nearly 200 Israeli sorties attacked about 100 targets and severely degraded Iran's integrated air-defense system. CSIS on air-superiority lessons from Iran and Ukraine
The Israeli Ministry of Defense says the campaign drew on more than 12,000 satellite images, extensive AI and electronic warfare, and a UAV force that performed more than 500 strikes and interdictions inside Iran. These are official Israeli claims and describe the campaign from the attacker's perspective. Their architectural meaning is still clear: sensing, software and strike had become one continuous activity. Israeli Ministry of Defense assessment of Operation Rising Lion
Iran was also developing the hunter for an emitting defense. An Air University analysis describes Iranian Kian drones designed to carry radar-wave receivers and home towards air-defense emissions. It concludes that Iran's 2025 suppression effort did not overturn the campaign, partly because mobile radars, decoys and electronic warfare diluted it. The direction of travel nevertheless matters: inexpensive uncrewed aircraft increasingly search for the sensor before another weapon searches for the city. Air University on suppression of air defense in Iranian doctrine
A sensor web answers this threat through abundance. It offers no single throat to cut. A microphone emits nothing. A passive radio receiver can resemble ordinary communications equipment. Cameras can move among civilian and military sites. Active radars can transmit briefly, relocate and leave behind other sensors that preserve the track.
The lesson from Iran is not that radar failed. It is that survival begins before the radar switches on.
04 — The city already contains the transmitters of a passive radar.
Passive radar begins with a strange proposition: allow somebody else to illuminate the target.
A television mast, FM station, Wi-Fi router or mobile base station continually fills part of the environment with radio energy. An aircraft crossing that energy changes a minute portion of it. Place a synchronized receiver somewhere else, compare the direct transmission with its delayed reflection and the receiver can infer that an object moved through the space between them. The city continues making phone calls and broadcasting programmes. The defensive system studies the shadows those signals cast.
A 2025 academic review describes these transmissions as illuminators of opportunity. Passive radar avoids a dedicated high-power transmitter, reduces hardware cost and makes the receiver difficult to locate through its own emissions. Experiments have detected or tracked small aircraft using Wi-Fi, 3G, 4G, digital television, satellite navigation signals and other broadcasts. The same literature identifies the remaining work: suppressing urban clutter, synchronizing separated receivers and preventing an algorithm from filtering out the tiny target together with the background. Academic review of passive radar for UAV detection
Fifth-generation mobile networks make the idea especially attractive. Their dense base stations, precise timing and higher-frequency signals create an existing geometry of transmitters around cities. An IEEE experiment has already examined a working 5G network as the illuminator for drone detection. IEEE study of 5G network-based passive radar
Passive radio sensing cannot detect every threat. A quiet autonomous drone may transmit nothing, while reflections vary with shape, angle and frequency. The network therefore gives cellular reflections a supporting role beside sound, heat, imagery and active radar.
The remarkable change is infrastructural. A future city will not need a military transmitter on every roof. Much of the radio light is already there.
05 — AI will turn weak clues into a track.
No individual sensor wins this forecast. Fusion does.
A microphone hears through darkness but struggles with wind, traffic and changing engines. A daylight camera can classify an object but loses contrast in cloud and night. Thermal imaging restores part of the night and loses detail in difficult atmospheric conditions. Passive RF can identify a transmission and fails when an autonomous aircraft remains silent. Radar measures motion and range while confronting small reflections, clutter and an adversary listening for the radar itself.
The failures are complementary. Sound can direct a camera. The camera can separate a bird from the object producing the radar return. RF direction-finding can connect the flight to an operator or confirm a familiar aircraft family. Radar can add the range that the microphone lacks.
An experimental system reported in Advanced Intelligent Systems demonstrates the sequence in miniature. A microphone array first estimates a drone's direction. A gimbal points a narrow camera and lidar towards that sector. The optical instruments refine the observation into a three-dimensional track. The broad but rough sensor recruits the narrow but precise one. Acoustic and optical drone-tracking study
A wider 2024 review finds the same logic across radar, RF, acoustic, visible and thermal systems: each modality supplies information that another lacks, and multimodal fusion improves classification and resilience. State-of-the-art review of drone detection and sensor fusion Newer research is moving beyond pairs. The TRIDENT framework synchronizes audio, visual and RF data and deliberately trains on degraded inputs so the model does not collapse when one sensor performs badly. TRIDENT tri-modal detection research
The military model will perform this fusion across geography rather than inside one laboratory rig. It will ask whether two microphones, one thermal camera and a radar return occurred at times and positions consistent with the same flight path. It will maintain several hypotheses, attach uncertainty to each and direct another sensor towards the place where one more observation will settle the question.
AI will not make the sky certain. It will decide which uncertainty deserves the next look.
06 — Intelligence at the edge will let the web grow without drowning its operators.
Twenty thousand cameras and microphones can create twenty thousand streams of distraction. The national sensor web becomes possible only when most raw data remains near the sensor.
A small processor beside a microphone can classify a sound locally and send a compact report: probable one-way attack drone, bearing northwest, confidence rising. A camera can transmit a track and a few identifying features rather than continuous high-resolution video. The command network receives objects, times and confidence estimates instead of an ocean of sound and pictures.
Britain's Defence Science and Technology Laboratory has formalized this principle in SAPIENT, an open architecture for autonomous networked sensors. Its nodes make local detections and classifications, exchange standardized messages and allow a decision module to correlate tracks and direct sensors. The UK Ministry of Defence has adopted SAPIENT as its counter-drone standard, and NATO exercises have connected sensors and command systems from different vendors through the same interface. A later move to a compact message format reduced required communications bandwidth by about 60%. UK Defence Science and Technology Laboratory on SAPIENT
This architecture makes attrition manageable. Destroy one node and neighbouring nodes retain their local models. Break one communications route and small messages can travel through another. Replace a camera or acoustic array with equipment from a different supplier and its reports can enter the same track engine. The system improves through software while its physical components remain varied and disposable.
It also creates obligations. A hostile actor will try to inject false tracks, imitate engine sounds, compromise cheap devices or flood the network with decoys. Each report will require authentication, time synchronization and a history of sensor reliability. In civilian territory, radio-frequency sensing must distinguish the physical character of a transmission from its private content. A 2026 US Department of Defense guide explicitly separates signal fingerprinting from prohibited interception and places radar, optical, acoustic and RF sensing inside one layered detection model. US Department of Defense guide to counter-UAS sensing and privacy
The web will scale because it sends conclusions instead of feeds.
07 — Better detection will protect the scarce interceptor.
An air-defense network does more than decide what to shoot. It decides what deserves an expensive shot.
Israel's Ministry of Defense says its layered systems intercepted 86% of ballistic missiles and more than 99% of Iranian drones during Operation Rising Lion. Even that performance allowed damaging missiles through, and it required a deep inventory of Arrow, David's Sling, Iron Dome and allied interceptors. The official figures describe extraordinary protection and an unavoidable remainder. Israeli Ministry of Defense assessment of Operation Rising Lion
The ammunition constraint sharpened during the renewed 2026 fighting. A July CSIS estimate placed US inventories below 1,000 Patriot interceptors and near 250 THAAD interceptors. It warned that diminished stocks could force defenders to accept more risk because there are few substitutes for those weapons against ballistic missiles. CSIS analysis of diminished interceptor inventories
A microphone web cannot replace the satellite and long-range radar that detect a ballistic launch. A passive cellular receiver cannot guide an exo-atmospheric kill vehicle. The sensor web contributes elsewhere in the decision: confirming low-altitude drones and cruise missiles, rejecting birds and false reports, recognizing decoys, predicting approach corridors and handing each target to an appropriate defensive layer.
A slow drone may receive an interceptor drone, automated gun or electronic attack. A cruise missile may justify a surface-to-air missile. A ballistic trajectory towards empty ground may require no engagement; the same trajectory towards a city may command the rarest interceptor. The shared track lets the command system make that allocation earlier and revise it as the evidence improves.
The future defense will still need costly missiles. It will spend them with a much richer understanding of what is coming.
08 — By 2032, the network will become the weapon system.
All the necessary pieces now exist separately. Ukraine operates a national acoustic layer under continuous attack. Britain has standardized the language through which autonomous sensors exchange reports. Academic teams have demonstrated acoustic-optical cueing, multimodal classification and passive radar using existing communications. Israel has shown how satellites, AI, electronic warfare and uncrewed aircraft can be integrated across an entire air campaign. Iran has shown why an emitting sensor and a finite interceptor magazine attract pressure from the first hour.
The next six years will connect those pieces. Front-line European states will extend passive warning across borders and critical infrastructure. Israel and Gulf states will spread low-signature sensors around the fixed sites that Iranian missiles and drones already target. Military radios will carry compact tracks produced by edge models. Telecom operators will discover that infrastructure built to carry conversation can also illuminate the air above it. Active radars will move, transmit selectively and inherit tracks that began as sound or borrowed radio light.
The first systems will produce false alarms. Their databases will be poisoned and repaired. Rival suppliers will resist common standards until operational necessity forces them to connect. None of these frictions alters the destination because the alternative is a shrinking number of conspicuous sensors attempting to classify an expanding number of cheap, adaptive threats.
By 2032, at least three countries beyond Ukraine will have crossed the line from installing counter-drone sensors around individual bases to operating a national or theater-wide web. Their public diagrams may still place the radar at the centre. Their software logs will tell the deeper truth: the track existed before the radar saw it.
The decisive anti-aircraft system of the next decade will not be a single machine. It will be the country learning to sense as one.
Loading interventions…