Radar’s pretty good at finding boats. Aim one at a harbor entrance and you’ll see the trawlers heading out, the ferry coming in and the guy in the dinghy who probably shouldn’t be out there after dark. What radar can’t do is look through water. So a diver, or a small drone swimming a few feet under the surface, just doesn’t exist as far as it’s concerned.<\/p>\n
That’s a bit of a problem when you think about what goes through a port. A lot of what’s in your kitchen and your gas tank came off a ship, and plenty of ports have a naval base parked inside the breakwater too. Below the waterline, most of them aren’t watched by much of anything.<\/p>\n
So how do you guard what’s under the water?<\/p>\n
You listen to it, basically. A German company called EvoLogics spent nearly three weeks trying exactly that in Sesimbra, a working fishing port on the Atlantic coast south of Lisbon. It brought underwater microphones, sonar, cameras and a fiber-optic cable laid on the seabed. Then a red team spent two nights trying to sneak past all of it.<\/p>\n
The setting was REPMUS, the exercise the Portuguese Navy runs every year for drones that fly, float and dive. NATO’s own wrap-up<\/a> says this edition ran from August 31 to September 25. It drew more than 3,700 participants from 36 navies, along with some 300 uncrewed systems. EvoLogics worked in the exercise’s maritime security area, which Portugal’s National Maritime Authority ran, and protecting a harbor was one of the scenarios on its list.<\/p>\n EvoLogics, which is based in Berlin, set up three listening stations on shore and two sensor buoys out on the water. Each one carries an array of hydrophones, which are just microphones built to work underwater. They’re passive, meaning they don’t send out any sound of their own, so whatever’s sneaking in has no way of knowing it’s being heard.<\/p>\n One hydrophone array on its own can only tell you which direction a noise came from. Put a few of them in different spots and you can work out where the noise actually is. Picture two people on different street corners, both pointing at the same car alarm. Where their arms cross is where the car is. That’s pretty much the whole trick, just with propeller noise instead of a Honda that won’t shut up.<\/p>\n It isn’t all quiet listening, though. The network also has an active sonar, which sends out pulses and times the echoes when somebody needs an exact distance on top of a direction. Then there’s the cable, a strand of fiber optics on the seabed that EvoLogics says picks up vibrations along its entire length, so the glass itself works like one very long microphone.<\/p>\n Above the water, day and thermal cameras check what the hydrophones think they heard, and software tags boats in the video on its own. The system also reads AIS, the transponder signal most commercial ships broadcast with their name and position. A trawler broadcasting AIS is basically wearing a name tag, and the boat that isn’t wearing one is the one you want to look at.<\/p>\n Everything runs over fiber and radio to a server in a container on the pier. Operators there see sonar, video and ship traffic side by side, sort each contact and pass it up the chain in NATO’s standard reporting format.<\/p>\n The red team’s job was to get into the harbor starting at least five nautical miles out, which works out to about 5.75 regular miles. It had combat divers, underwater drones, drone boats and small fast boats to work with, and it came in both on and below the surface.<\/p>\n The first night was close to ideal. The water was calm, visibility was good and background noise stayed moderate, which is about as easy as a harbor gets. According to EvoLogics, all three shore stations heard the underwater drones once they were inside the harbor, and the bigger ones got picked up early, while they were still on their way in.<\/p>\n Then the weather turned. A strong wind out of the northeast pushed waves into the breakwaters, and fishing boats kept coming and going, flooding the sonars with noise. That’s the night I’d pay attention to, frankly, because a real harbor isn’t going to send its fishing fleet home so the sensors can hear better.<\/p>\n EvoLogics says the network kept tracking its targets anyway. One underwater drone was followed continuously inside the harbor up to the mid-raid break. Drone boats were detected and logged, and a small open boat with two people aboard stayed on the thermal camera at the harbor entrance. One of those drone boats was classified and cleared for reporting 30 seconds after the system first picked it up.<\/p>\n Thirty seconds sounds great. I’d just be careful with it.<\/p>\n That’s one drone boat on one night, and EvoLogics doesn’t present it as a typical time for everything the network caught. It also doesn’t say how far away that boat was when it got picked up. And that five-mile figure is where the red team started, not a range for the sensors.<\/p>\n Maybe. The red team’s divers went in from a nearby beach, and within 20 minutes the seabed cable registered signals and a contact was reported. The detection software flagged a possible diver near the harbor entrance.<\/p>\n EvoLogics hasn’t called that a catch, and to its credit, it says so plainly. The company is still going through the data to confirm the contact and rule out a false alarm. So for now, you should read it as a promising blip, not a diver in handcuffs. I’d honestly rather see a company admit that than oversell it, because divers are arguably the hardest thing on this list to hear. A person finning slowly along doesn’t make anything like the racket a propeller does.<\/p>\n The most convincing result might not have been on the schedule at all. During the trial, a fishing boat hit one of the network’s own sensor buoys and damaged it.<\/p>\n The EvoLogics team then rebuilt the collision from what the system had recorded: the acoustic signatures, the sonar tracks, the AIS data and the video. The company pitches that as a use for accident investigations and maritime police work, on top of the security job.<\/p>\n And I think that’s the better pitch, to be honest. A port might go years without a hostile diver showing up, but boats bump into things all the time, and somebody always wants to know whose fault it was. I suspect a few harbor masters would pay for that feature alone. EvoLogics hasn’t said how badly the buoy was damaged or which boat hit it.<\/p>\n Every result above comes from EvoLogics’ own account of the trial<\/a>, and the trade press coverage I’ve seen repeats that same account. NATO’s summary of REPMUS sticks to the big picture and doesn’t get into the harbor scenario. That’s pretty standard for exercises like this, as far as I can tell, and it doesn’t make the results wrong.<\/p>\n There are still a few things I’d want to see before getting too excited. The company says its sensors threw up hundreds of alerts between September 7 and 17 alone, but it doesn’t say how many were real intruders and how many were a trawler clanking past. It doesn’t publish detection ranges, and as far as I can tell, it hasn’t named a customer yet.<\/p>\n If you’ve been following our coverage, the basic idea will look familiar. This summer, an American seabed drone heard underwater drones come through a channel<\/a> and put them on a Navy screen within seconds, and we also went through why quiet listening and active pinging work better as a pair<\/a>. EvoLogics is running a similar playbook from fixed stations around one busy fishing port.<\/p>\n Finding an underwater drone is only half the job, of course. Stopping one is a whole other problem, which is why a German sonar company built a 24-pound mini-torpedo<\/a> just for that and brought it to the same exercise.<\/p>\n EvoLogics published its account of the Sesimbra trial on October 9, and the Berlin company says the harbor network will be available next year.<\/p>\n","protected":false},"excerpt":{"rendered":" Radar’s pretty good at finding boats. Aim one at a harbor entrance and you’ll see the trawlers heading out, the … <\/p>\nIt’s three shore stations, two buoys and a cable on the seabed<\/h2>\n
Night two was the real test<\/h2>\n
So did it catch a diver?<\/h2>\n
A fishing boat ran into the test<\/h2>\n
So far, these are EvoLogics’ own numbers<\/h2>\n