I understand the physics behind how the cables themselves work in transmitting light. What I don’t quite understand is how it’s possible to convert millions of messages, emails, etc every second and transmit them back and forth using only a few of those transoceanic cables. Basically, how do they funnel down all that data into several cables?
In: Engineering
A key feature of fiber optics is the ability to multiplex multiple wavelengths of light onto the same strand of glass. Think of it like a prism- the one beam of white light can be broken down into multiple beams of colored light. Each of those “colors” (the light used in fiber operates at wavelengths invisible to humans) can carry as much data as your modulation scheme allows for- say 1Gbps to make it easy.
Using dense wavelength division multiplexing (DWDM) we can fit 80 of those wavelengths onto one strand. And of course most fiber cables are multiple strands- lets say its a small cable with only 12. Doing the math, that’s almost 1Tbps of bandwidth on a fiber cable.
Real life applications can carry even more, it all depends on the equipment being used. As our technology improves we’ll be able to modulate more data per wavelength and pack more wavelengths per strand.
Really the answer is Packets. The same way you can have an ethernet connection and get e-mail, be watching a video on a web browser and playing an online game all the the same IP address. It happens the same on the transport end except a much larger bandwidth so more things can happen at the same time.,
1. The cables are huge, and there’s a lot of them
2. They use fiber optics, which means they communicate through light. Light can have different colors, so if you use 6 colors instead of 1 color, you multiplied your bandwidth by 6.
Gee, everyone talking about DWDM and stuff when that is really almost irrelevant to the question. Yeah, sure, having n wavelengths on a fiber reduces the number of fibers you need by a factor of n, which is great, but if you have hundreds of millions of “connections”, it having a thousand “channels” rather than a hundred “channels” does not really answer the question.
**The real answer is: Packets!**
What you perceive as a “connection” of sorts, at the network level, just isn’t. You might be thinking of how traditional telephone networks worked, where a pair of copper wires was (more or less) connected up between caller and callee to establilsh a connection. That is not how modern computer networks work. Modern computer networks are what is called **packet switched** (as opposed to **circuit switched**).
**As far as the network is concerned, there simply are no connections.** In the particular case of IP (Internet protocol–the thing you use for websites and email and instant messaging and what have you …), how that works is that every device connected to the internet has a unique number identifying that device, also known as the IP address. And that really means every device, your laptop or smartphone just as any one of Google’s or reddit’s servers. Now, when you(/your device) wants to send some piece of data to any other device on the internet, all it has to do is to label it with the address of the device it wants to send it to (that’s then called a packet) and transmit it to its upstream internet router. That router will look at the address to determine which of its available links (“cables”) would be the best (fastest/cheapest/whatever) choice to get the packet to the device that has that address. At the other end of that link, there will usually be another router that does the same thing. And then another one. And another one. … until, at some point, it reaches the router that the destination device is connected to, so the router will transmit it to that device, which will then, presumably, somehow do something useful with it–and also, oftentimes, send another packet in reply. For that to work, the sender of a packet also adds its own address to any packet it sends, so the recipient can use that address to send back a response.
However, the size of packets on the internet is limited, commonly to around 1500 bytes. So, if you want to transmit something that is larger than that, what you have to do is to split it up into small pieces and transmit those as individual packets–and you have to add some information to the packets that allows the receiver to put them back together, of course. But the important part here is that, as far as all those routers are concerned, there is no connection. They see a packet with a destination address, select a next-hop link to transmit it to, and forget about it. If you do some hour-long download, the routers don’t know and don’t care, all they see is individual packets, millions of them. And really, there is no reason all those packets would even necessarily take the same path through the same cables or routers. All that matters is that all those packets get to their destination, somehow. Or, well, most of them at least, because even that isn’t guaranteed, some fraction of packets do just get dropped for various reasons, in which case the sender will simply have to retransmit if they don’t hear back from the recipient in time.
So, the answer to the question is that those fiber links simply have an extremely high speed, and the routers connected to them simply push through them any packets that arrive in whatever order they arrive in. If your submarine cable has a capacity of 100 Terabit/second, say, that means it is capable of transmitting about 8 billion packets of the typical maximum size per second–and as far as the routers and cables are concerned, those could all be from the same “connection”, or they could be 8 billion different “connections”. In practice, usually, there would be a few million “connections” using that link during any second: Some people transfering data between datacenters or if you have a gigabit connection at home, you might be transferring ~ 80000 packets per second, someone else using a dialup modem in a rural area might be transferring only ~ 4 packets per second. All of those are just mixed as they arrive and transmitted through the cable, one after the other.
So many answers here, but none of them are explained like you’re five. I’m sorry OP. Here’s an actual explanation to answer your question:
The cables are really big.
I’d like to know how they are laid? Ship has a giant reel on the back laying thousands of miles of cable? How does that work?
In the earliest days of wire-based communication (analogue), there was only one signal on the wire at a time.
Later, we learned to pass multiple signals by using different frequencies/wavelengths for different signals. In it’s simplest forms, imagine that instead of shining a white light down a wire for one signal, you’re shining red, green, and blue lights. By measuring the amount of red, green, and blue light at the other end, you can separate the signals. That’s the principle, just scaled up and with a bit of fiddly detail at each end. This gives us (with current technology) up to about 80 signals per strand of fiber. A cable can have dozens of fibers, so that’s potentially a few thousand signals in a cable with a hundred fiber strands
But as you say, there are millions of data transfers… this part is fairly easy though. Although the signals appear to be fully simultaneous, we don’t actually have a constant connection along each possible pathway: what we do is break our data up into chunks called “packets”, and put an address on it of where it needs to go (your IP address).
That might sound strange, but think of it like sending things in the mail. Instead of having someone drive back and forth and give the message, we put them in packets, and send them down the road with an address on it.
Think of our cable as a road network. We make a road (a strand of fiber) and then we give it 80 lanes: those are our multiple signals. Now imagine we can overlap those signals so all 80 lanes are just one lane where the cars don’t interfere with each other.
And then for each lane, we can send lots of individual cars (packets) down one after another, with their own destinations. We don’t need to make a full connection (lane) between each pair of destinations, we just send everything down the shared lanes and let it split off when it needs to.
And then we combine a whole bunch of these roads (fibers) into one cable. So imagine we have 100 of these roads next to each other.
Imagine the cable fiber communication works similar to how Morse code lines work. They send data via lights on and lights off (0 and 1) instead of the regular _ . from Morse code.
Specialized machines (routers, but waaaay more powerful than the one you have at home) code and decode this 0-1 into usable date.
Remember that light travels incredibly quickly (about 0.01 seconds in a transatlantic fiber cable, example Yellow/AC-2), so the 0 and 1 make it from one side to the other really fast.
Now imagine that on top of this you can use red for one line of messaging and blue for another, and yellow, green, etc. So that you can use several lines of communication for each fiber.
And also note that Everytime they make a transatlantic cable they use a LOT of single fibers because if you are on the bottom of the ocean puting one fiber or 100 is almost the same cost.
That’s how you get data speeds of up to 160 terabits per second across the ocean (see MAREA transatlantic communication cable)
Hi! Here is a good link ( [https://twistedsifter.com/2012/07/the-undersea-cables-that-connect-the-world/](https://twistedsifter.com/2012/07/the-undersea-cables-that-connect-the-world/) ) on what the fiber looks like. It is super durable because, well, there is a lot of things that can go wrong at the bottom of the ocean. In essence, there are normally always one or two cable ships laying undersea cable at any given time, sometimes to fix lines that broke and sometimes to give more bandwidth. From what I have read before, to solve attenuation every 100KM or so there is a repeater device.
There isn’t only one or two wires in a cable bundle, there can be hundreds. The cable is huge, but the wires themselves are the width of a human hair. I can’t find the exact number of wires in the undersea cable but from looking at the picture they must have at least 48 pair.
So, quick explainer, on the internet you typically have one wire for send and one wire for receive. So if you want a 10Gb connection between two pieces of electronics, you would need two wires. So if you have a 48 wire cable, then you get 24 different send and receive pairs that can be put into electronics.
Technology has come along and we know how prisms work. For example, I have residential F/O service in my house. They gave me 1 wire. So how do I send and receive? The technical answer is ‘coarse and dense wave multiplexing’ which is a complicated way of saying you can send a communication in one direction on one wavelength (or ‘color’ for short) and receive on a different wavelength. The colors are a handy way of talking about it, but these electronics can use the non-visible spectrum as well. I think, the last time I purchased multiplexing optics, I was able to get 10 10Gb pairs on one wire. So for that undersea cable, assuming it is 48 pin, it is 48 x 10 send/receive pairs.
How the data actually flows? That is a topic called ‘digital signal processing’ (DSP) and you will need to ask another ELI5 for that.
You mentioned “only a few of those transoceanic cables” — there are far, far more than a few. There are a few hundred, some longer than others but all serving the same purpose.
Latest Answers