How does that work in practice, anyways? I mean, cables these days don't just carry a bunch of fibre wires, but also high-voltage cabling to provide power to the repeaters. Isn't it risky / detectable to cut open such a cable?

Also, how is the spliced data actually exfiltrated back to land? The obvious candidate is either a leased dark fiber (which requires intimate knowledge of how the individual fibres are arranged) and laying a second cable (with the obvious disadvantage that cable-laying ships are rare and can easily be spotted).

>how is the spliced data actually exfiltrated back to land?

During the Cold War, they periodically returned to the site of the tap and changed the tapes and batteries. One of the taps was installed on a Soviet telephone cable across the Sea of Okhotsk in several hundred feet of water. It was a 20-foot long contraption able to eavesdrop without penetrating the cable. It was designed to fall off the cable if the Soviets raised it for repairs. It survived nearly a decade, until 1981, when Ronald Pelton sold its location (among other secrets) to the KGB for $35k.

http://www.military.com/Content/MoreContent1/?file=cw_f_ivyb... https://en.wikipedia.org/wiki/Operation_Ivy_Bells

An inflation calculator says it is $100k today. That sounds awfully cheap.

The cut is made 100's of miles away from where the taps are being installed and made to look accidental because installing the taps will disrupt service. When the technicians run the OTDR, a repair ship is sent and the distraction cut is repaired, but the taps have already been installed and sealed back up. Not all fibers pass through places where it can be done on land (i.e. not in US) so this is necessary.

submarine cables are significantly different than terrestrial fiber networks. a fault will be detected in/from both directions, immediately and automatically.

it has been suggested that the USS Jimmy Carter can tap submarine cables without disrupting service or physically damaging the fiber.

did you know when you bend a strand of fiber enough, you can make out the signal and only affect attenuation?

Yeah but still you need to remove the outer casing in order to, well, see the fiber. And that is detectable if you have to cut through a layer conducting high voltage.

Modern cables don't have high voltage to worry about. The signals are amplified optically:

https://en.wikipedia.org/wiki/Optical_amplifier#Doped_fiber_...

It isn't necessary to bend a fiber either. Optical fibers can be tapped via evanescent wave coupling. The fiber cores just need to be near enough to each other over a long enough span, possibly with some removal of the outer cladding.

Notable patents:

"There is a need for coupling light into and from an optical transmission line without interrupting the flow of transmission. Such lines may be used for communications, as well as other purposes.[1]"

1. https://www.google.com/patents/US3933410

2. https://www.google.com/patents/US4097117

3. https://www.google.com/patents/US4377403

4. https://www.google.com/patents/US4763272

5. https://www.google.com/patents/US4336047

Raman amplifiers still require juice to run the pumping laser. There aren't exactly 110V sockets sitting on the ocean floor.

There is a copper tube around the fiber bundle that carries high voltage (usually 10,000+ volts to reduce losses over long spans). The return is usually just coupled directly to the ocean.

If you're careful, use a non-conductive cutting blade, and have enormous balls, you can no doubt cut through the tube on one side and not cause any noticeable effects. At least, if you've brought along someplace dry to work on it...

Yeah it's just not that easy with the advent of coherent optical communications in subsea networks. You need to be able to do some very complex DSP in real-time at 50Gb/s. You also need to know the exact FEC encoding scheme, grey map, actual modulation format etc... It is very non-trivial. For OOK signals I think it's somewhat doable for coherent, very, very hard. Not totally impossible but I highly doubt even CIA has this expertise.

"Where there's a will (and enough black project money), there's a way."

With love, NSA

Submarine cable is nitrogen doped to prevent hydrogen intrusion.

Would make sense to just tap into the repeaters. That is the common sense approach from someone (me) that knows zip about the setup but that seems like the way it would be done.

They're optical repeaters.

> Also, how is the spliced data actually exfiltrated back to land?

In the operations against the Soviets that've been made public, they physically retrieved the tap.

Why bother when they have direct access to the US end of it anyway?

Why that submarine in particular?

Supposedly it has a cable splicing room.

http://www.nytimes.com/2005/02/20/politics/new-nuclear-sub-i...

Because the USS Parche (SSN-683) is decommissioned?

my thoughts exactly.

my friends and I always half joke about this too. It's likely the sad truth.