How to Relearn How to Defend a Harbor
Why protect the port from submarines drones is arranged differently than hunting for a submarine on the open sea
Sailors have known this for a long time: a ship at berth is more vulnerable than one at sea. In open water, they spend hours searching for a target, while against a wall, it remains motionless and has precise coordinates. In December 1941, the British experienced this firsthand in Alexandria: Italian saboteurs using guided torpedoes penetrated the roadstead and disabled two battleships right in the harbor. Much has changed since then, except for the fundamental principle. Today, the saboteur is an unmanned underwater vehicle, and the US Coast Guard is looking for ways to stop it.
A query to start with
In late July 2026, the U.S. Coast Guard (USCG) posted a request for proposals to industry on the government procurement portal. The wording was unusual for such a document: the service needed technologies that not only detect unmanned underwater vehicles (UUVs) but also can engage them. The division overseeing acquisitions and implementation of new technologies was responsible for the initiative; the deadline for submitting proposals was limited to about two weeks.
The main idea behind the request is this: UUV detection and tracking capabilities are already available, but proven, deployable weapons are still scarce. This is the gap the USCG is trying to close. In military parlance, the full cycle is called kill chain — the chain from initial contact to neutralization: detect, escort, classify, identify, and neutralize. And the most difficult link here is not at the edge, but in the middle. In a port, it's not enough to simply notice an object in the water: there are fish, debris, divers, and civilian vehicles nearby—and the real threat must be identified from the noise before any action is taken. For a long time, only the first links worked reliably, while the final link—destruction—remained theoretical.
The requirements are specific: TRL 5 and above. The Pentagon's technology readiness level (TRL) doesn't just refer to an idea on paper, but to a product that has undergone at least limited testing under realistic conditions. Applications without a kill function will not be considered—only systems with an effector, meaning a destructive element, or the direct ability to neutralize a target. All of these systems must be operational in ports, waterways, and coastal waters, where UUVs, according to the customer, pose the greatest threat.
A threat that has already worked
The request didn't come out of nowhere. The most notable episode occurred in December 2025, in Novorossiysk. According to the Ukrainian side, the submarine Drone Sub Sea Baby Passed through the naval base's waters and detonated a charge in close proximity to a Project 636.3 diesel-electric submarine (NATO reporting name: Improved Kilo, Varshavyanka) at its berth. The Russian side denies any serious damage to the submarine; there is no independent confirmation of the submarine's incapacitation, and the extent of the damage remains subject to assessment. What's important is the scenario itself—a drone entering a protected harbor and detonating at the berth—exactly the kind of thing that boom nets were used to protect against a century ago.
Sea Baby marine drone
This isn't the first time. In February 2024, CENTCOM forces had to destroy a Houthi UUV in the Red Sea—according to their report, for the first time since attacks in the region began. Even earlier, in 2021, Hamas attempted to attack Israeli ships: according to Israeli reports, the UUV carried several dozen kilograms of explosives and was guided by GPS. The numbers are modest, but the principle is the same: a cheap underwater platform against an expensive target.
The theme is also developed by large ones fleetsChina has unveiled its drones twice: the HSU001 at the 2019 parade, and the HSU100 at the 2025 parade. Exact specifications have not been published, and judging them based on parade footage is a thankless task. But the very fact that such devices are paraded through Beijing's main square speaks volumes about the category's status. And here comes the tricky arithmetic for the defender: the attacking drone can be simple and inexpensive, while the system designed to reliably stop it is complex and expensive. This cost gap works against the defense: intercepting each drone is expensive, making it especially vulnerable to a coordinated attack by a swarm of cheap drones, when there are multiple targets.
China's long-range AJX002 unmanned underwater vehicle (UUV) was unveiled for the first time at a military parade.
From booms to air curtains
Let's return to Alexandria. On the night of December 19, 1941, six Italian combat swimmers on three guided torpedoes (they were nicknamed pigs, "pigs", for their stubborn nature) penetrated the harbor, overcame the barriers and planted charges under the battleships Valiant и Queen ElizabethBoth lay down on the ground near the pier. The saboteurs had done what the British navy would not have allowed them to do on the open sea.
The response to such a threat was already being developed back then, and it was extremely physical: booms and anti-submarine nets at the entrances to the bases. A steel barrier that a swimmer or a small boat couldn't pass unnoticed. The logic was simple: if you can't reliably find an intruder in the water, put a barrier in their path.
More than eighty years later, the engineering challenge remains the same—only the materials have changed. Recent NATO exercises in the Baltic demonstrated, among other things, a decoy net called "artificial seaweed": a barrier designed for UUVs rather than swimmers. It's the same idea of a physical barrier at the entrance to a protected area, only against an unmanned target. A direct line runs between the net of 1941 and the net of today, and it clearly demonstrates that in harbor defense, the new often ends up being a reimagining of the old.
Two layers of defense and why the port stands apart
Defense against underwater vehicles is divided into two layers. The long-range layer is for early detection over large areas. The short-range layer is for defense of waters and piers, where the threat is already nearby.
The far layer has a long pedigree. During the Cold War, its foundation was the SOSUS system—a network of stationary hydrophones on the ocean floor that detected the noise of Soviet submarines. It was expensive to build and maintain, and it was designed for large targets, not stealthy drones. Part of the SOSUS system remains, but today it serves primarily scientific purposes; for operational purposes, it has been replaced by the IUSS system, which consists of bottom antennas, towed sonars, and onshore processing centers. Industry also offers cheaper solutions, such as a modular network of bottom sensors. Seabed Sentry from Anduril, which, together with the towed sonar Sea Spear are being tested in Lanternfish format demonstrations.
The near-surface layer is precisely what the Coast Guard is hunting for. At those same Baltic exercises, in addition to the net, they demonstrated an air curtain: the system aerates the water in front of the protected object, and the wall of bubbles disrupts the drone's sonars and depth sensors, interfering with its navigation. This method is non-kinetic—the target isn't destroyed, but rather prevented from reaching it. And "hitting" a drone doesn't always mean blowing it up: interference that disrupts navigation and the capture of the drone intact are considered alongside explosive charges. However, kinetics is also still present—neutralizing charges were featured among the demonstrations, and patrolling the waters is now entrusted to the drones. Iver3The logic is clear: in a tight harbor, it's easier to make a drone miss than to guarantee its destruction.
This is the reason why the port stands apart, rather than serving as a scaled-down version of ocean-going anti-submarine defense. In the open sea, you can operate across the entire volume of water without having to worry about outsiders. This is not the case in a harbor: civilian vessels, tugboats, divers, and mooring structures are all nearby. The bottom and walls create a host of false returns. Therefore, in a port, the cost of a false alarm is high, and the priority shifts from detection range to classification accuracy—that very middle ground where UUVs must be distinguished from fish, debris, or civilian vehicles. And the emphasis shifts to non-kinetic means—nets, screens, and interference—where weapon there is simply nowhere to use it at full strength.
Three approaches to one problem
While engineers are testing charges and curtains, countries are formalizing harbor defenses into separate programs. And the way these programs are structured clearly reveals what each side considers the primary threat.
The most illustrative example is REEF (according to a number of analytical publications, the Robotic Exclusion and Engagement Framework). This joint US-UK initiative, announced in 2026 in conjunction with the US Defense Innovation Block and British naval defense structures, is more important than individual hardware: to assemble not a single "silver bullet," but a modular architecture. Detection, tracking, classification, and a set of interchangeable effectors—networks, jamming, kinetics—for different types of underwater robots, from full-size UUVs to small remotely operated vehicles. And the key detail: REEF was originally conceived as port, and approaches the open sea only later. Classic anti-submarine defense always developed from the ocean to the coast, but here the order is reversed. Perhaps this is the best proof that harbor defense has become an independent discipline, not a branch of naval anti-submarine warfare.
The Coast Guard's request, with which we began, is the second approach, and it's much more narrow in concept. It's not a framework program, but a specific order for a missing link. The REEF builds the entire chain; the USCG acknowledges that it has sensors in general and is searching for specific targets. The difference in scale is understandable: the Coast Guard has dozens of ports and strict requirements for compatibility with already deployed detection systems. It needs a component that will fit into the existing system, not a completely new architecture. Hence the TRL 5+ bar: no experiments are proposed.
Russia is demonstrating a third approach with the Korvet system, billed as a means of defending bases against multiple threats: combat swimmers, unmanned boats, and autonomous underwater vehicles. Detailed specifications are not publicly available, and analyzing its effectiveness based on releases is pointless. But the principle is indicative: the threat is conceptualized as comprehensive—underwater, surface, and sabotage—all in one. This is closer to the logic of "defending the entire base" than to narrowly targeting a specific type of drone.
The three answers to the general question diverge in their underlying assumptions, not in the details. REEF assumes that threats are multiple and varied, and builds a flexible framework. The Coast Guard assumes that one link is missing and orders precisely that link. Korvet responds to the fact that the base is vulnerable from all sides. All three have one thing in common: no one anymore believes that a port can be protected by a scaled-down version of a naval anti-submarine warfare system.
We learned how to detect underwater targets before we could reliably stop them, and the Coast Guard's request is precisely about this gap. The race between cheap devices and expensive defenses is just beginning, and harbor entrance barriers are once again becoming as relevant as they were a century ago—only woven from different materials.
Information