光通信 - 第1集 - 光线与波

今天，我们要讨论的是，你现在正在听的内容，正通过一些直径仅为人类头发十分之一的玻璃纤维传输到你的耳朵里。

Today, we're talking about the fact that what you're listening to right now is reaching your ears, partly taking the route through glass fibers that are the tenth of the width of of a human hair.

除非，

Unless, of

当然，如果你是我邻居，那你就是透过墙壁在听。

course, you're my neighbor, then you're listening to it through the walls.

是的。

Yeah.

所以今天在《光波与电磁波》播客中，我们要讨论的是光通信，更具体地说是光纤通信。

So today on the Rays and Waves podcast, we're talking about optical communications, but more specifically fiber optic communications.

我们会带你们了解这项技术的起源、目前的发展，以及所涉及的一些技术，还有这堪称人类工程学的一项惊人成就。

We wanna walk you through where this all began, where it's gotten to, and some of the technologies involved, and what's honestly quite an amazing feat of human, engineering.

对。

Yeah.

在过去十年中，光纤通信网络的建设构成了互联网的硬件基础。

In the last decade, the construction of the optical fiber communication network has has formed the hardware foundation to the Internet.

通过这样做，它极大地促进了人类生活方式各个方面的变革。

And by doing so, it has greatly contributed to, well, basically changing all aspects of how we humans live our lives.

是的。

Yeah.

我的意思是，已经部署了60亿公里的光纤。

I mean, 6,000,000,000 kilometers of fiber has been deployed.

作为参考，这大约是地球到太阳再返回地球距离的20倍。

As a frame of reference, that's around 20 times the distance to the sun and back to Earth.

如果你把这根光纤拉成一条直线，几乎可以直达太阳系的边缘。

You could basically reach the end of of our solar system if you drew that that strand of fiber out in one one go.

尤其是如果你是那些不再认为冥王星是行星的异见者之一的话。

Especially if you're, one of the heretics who no longer believes that Pluto is a planet.

那么你确实已经到达了太阳系的尽头。

Then then you're get definitely getting to the end of the solar system.

而根据你所包含的具体内容，2023年光通信市场的价值在100亿到300亿美元之间，未来十年的年均复合增长率约为10%。

And depending on kind of what you include in it, the optical communications market value is somewhere between 10 and 30,000,000,000 US dollars in 2023 with a cumulative annual growth rate of about 10% in the coming decade.

这主要得益于高速连接的普及。

And that's largely driven by the adoption of high speed connectivity.

目前，互联网流量以每年约20%的速度稳定增长，并预计在未来十年内继续保持这一趋势。

Currently, the Internet traffic has been steadily growing with a pace of about 20% per year, and is foreseen to take the same trend over the the coming decade as well.

但试图为光纤通信赋予市值时，有趣的是，这与对互联网整体的估值论证是相同的。

But the interesting thing about trying to attribute a market cap to fiber optic communications is it's the same argument that holds for the internet in general.

你可以尝试给它一个市值，在这个案例中，光纤通信的估值大约在100亿到300亿美元之间。

You can try and put a market cap on it, and you'll get some number in this case between 10 and 30,000,000,000 for fiber optic communications.

但它支撑着整个互联网，你可以合理地认为，现代社会在某种形式上都依赖于互联网。

But it underpins all of the Internet, and you can reasonably make the argument that modern society is in one form or another underpinned by the Internet.

因此，有估算认为它的市值在100亿到300亿美元之间，但同样合理的是，从某些衡量方式来看，它对经济的贡献接近100%甚至更高。

So it's a reasonable argument to make that by some estimates, it has a, market cap of 10 to 30,000,000,000, but it's also reasonable to say that its contribution to economics is, in some ways of measuring it, like, a 100% or very near to it.

是的。

Yeah.

我想，这100亿到300亿美元大致是建设光纤网络的上限，比如

I guess that 10 to 30 billions is kind of the cap for building your fibre, like,

那就是维护光纤。

it's that maintaining fibre.

是的。

Yeah.

因为它产生的价值在百分比上要大得多。

Because the value it produces is vastly as a percent.

是的。

Yeah.

是的。

Yeah.

它所实现的功能远远超出了它的收费。

What it enables is far beyond what it, charges.

为了了解当前光纤通信的速度水平，如果将世界分为两部分，分别位于单根光纤的两侧，理论上，一根光纤就能同时支持50亿通电话，让一半人口与另一半人口进行通话。

So just to get a feel for the state of the art of the speeds that you can now get with optical fiber communication, if you separate the world into two camps, one on either side of a single strand of optical fiber, you can, in principle, support 5,000,000,000 phone calls to half of the population talking to the other half at the same time with one single optical fiber.

但我们当初是从哪里开始的呢？

But where did we all start with this?

在我们能够实现每秒100太字节的光纤通信之前，早在古代就已经使用光学通信了。

Long before we were able to have a 100 terabyte optical fiber communication, we were actually using optical comms back in antiquity.

我在为这一集做研究时发现的一个例子让我印象深刻：特洛伊的陷落是通过数百公里长的一系列火信号传递给希腊人的。

One example that I found in researching for this episode that I really liked was that actually the fall of Troy was signaled to the Greeks through a series of fire beacons over hundreds of kilometers.

他们点燃火焰，让整个希腊语世界知道特洛伊已经沦陷。

So they would light fires to let the rest of the Greek speaking world know that Troy had fallen.

这无疑是一种非常原始的光学通信形式，但其基本原理并没有本质区别。

And this is, you know, a very rudimentary form of optical communications, obviously, but it's actually not fundamentally different.

即使是现代的光纤通信，你依然只是在开关灯光，只不过做得更快、更巧妙一些，但核心原理是一样的。

Even modern optical communications, you're still just turning lights on and off or just doing it faster and in some slightly more clever ways, but it's the same basic principle.

是的。

Yeah.

所以通过偏振和光的特性等手段玩一些花样。

So playing a bit of tricks with polarization and playing tricks with the face of light and so on.

但归根结底，我们依然在闪烁灯泡，通过这些光来解码信息。

But at the end of the day, we're we're still blinking lamps on and off and decoding the information that comes through that light.

当然，烟雾和火光信号是光学通信最初的原始尝试。

Smoke and fire signals, of course, was the first rudimentary foraying to optical communication.

但到了近代，电力尤其是格雷厄姆·贝尔发明的电话迅速占据了主导地位。

But quickly when getting to a bit more modern times, electricity and especially the invention of the telephone by Graham Bell took precedence.

然而，格雷厄姆·贝尔还发明了另一种基于光学的通信方式，他本人认为这是自己职业生涯中最伟大的成就，那就是光电话——你对着一个膜片说话，膜片产生信号使镜子变形，而镜子则收集阳光并将其投射到接收端。

However, Graham Bell invented another optically based communication method that he himself thought of as the crowning achievement of his invention career, which was the photophone, where you spoke into a membrane which generated a signal that deformed a mirror, and that mirror was collecting sunlight and directing that to the receiver.

根据镜子形状的变化，它会改变光束的发散角度，在接收端表现为光强的变化，而接收端的光电探测器会将这种变化转换为声音信号。

Depending on how the the mirror changed its shape, it changed the divergence of the lights picking up as a intensity variation in the receiving end, which which is the photo detector that was then converted to a a sound signal.

但到目前为止，我们还没有真正谈到光纤，而这才是我们答应要跟你聊的主题。

But none of this so far has actually been about fiber optics, which is what we promised to talk to you about.

那么我们现在就转到这个话题上。

So we'll switch over to that now.

光纤在本质上与特洛伊的火光信号，或者在某种程度上与光电话，是完全相同的东西。

Fiber optics is basically the same thing at its core as fire signals, back in Troy, or to a certain extent, the photophone.

光纤只是作为一种传输我们开关闪烁的光的手段，从而实现信息的传递。

The fiber optics just act as a way to transport light that we're blinking on and off so that we can send information through.

支撑这一原理的一个基本光学原理称为全内反射。

One of the things that underpins it is a fundamental principle in optics known as total internal reflection.

如果我们想象一个棱镜，将光以特定角度射入，它会穿过一个面，击中背面，然后反射到另一个面。

If we imagine a prism for instance, and we shine light into that prism at a certain angle, it'll transmit through the one face, hit the backside, and then be reflected out to the other face.

而当入射角超过某个特定角度时，光将被完全反射，不会有任何部分透射出去。

And now above a specific angle of incidence, the light will actually be completely reflected and none of it will be transmitted through.

因此，反射光束的强度将达到100%。

So you'll have a 100% intensity to the reflected beam.

所以，如果你想象一根光纤，将一束光射入其中，它会在内壁上反射，如果以很小的角度入射，光就会完全被限制在光纤内部。

So if you can imagine a tube of fiber that you're sending a ray of light into, it'll reflect off the sides and if it's coming in at a glancing angle, it'll all stay within the fiber.

通过这种方式，光可以在极长的距离中传输。

And in this way, it can transmit over incredibly long distances.

史蒂文刚才所描述的光线模型，对于建立直观理解非常出色，它适用于所谓的多模光纤，在这种光纤中，光可以沿着不同的路径传播。

The ray picture that Steven was just talking through is fantastic for building an intuition, and it's what's going on in so called multimode fibers where light is allowed to take different paths through the optical fiber.

然而，长距离光纤通信网络的主力是所谓的单模光纤。

However, the workhorse of the optical fiber communication long haul network is something called single mode fibers.

而在那里，光线模型实际上有点失效了，你不得不改用波动模型来理解，此时只允许单一的高斯模式光通过光纤传输。

And there, the ray picture actually kind of breaks down a bit and you're you're forced to view it in a wave picture instead where a single Gaussian mode of light is allowed to be transmitted through the fiber.

但要理解光通信网络，我们不需要深入探讨光线与波动的细节，或者模式在小型光学结构中为何是有限的。

But for an understanding of optical communication networks, we don't need to delve too far into the details of rays versus waves or how, modes are finite in small optical structures.

现在，我觉得有必要提一下，是因为之前谈到了光线和波动的播客内容。

Now, I figured it was worthwhile mentioning because of the rays and wave podcasts.

当然。

For sure.

我不是在反驳这一点。

Not, I'm not disputing it.

我只是想说，就这个播客而言，你真正需要知道的是，由于全内反射或模式约束，光可以在光纤中以极低的损耗传输极长的距离。

Just just saying that, for the sake of this, podcast, all you really need to know is that in optical fibers, because of total internal reflection or mode confinement, the light can transmit through it for incredibly long distances with low loss.

这正是光纤成为跨大西洋或跨太平洋数据传输强大工具的原因。

And that's what makes optical fibers such useful tools for, transmitting data across The Atlantic or across The Pacific.

我们提到过，这些光纤的直径大约为10微米，相当于人类头发直径的十分之一。

We mentioned that, these optical fibers are around 10 microns or tenth the width of a human hair.

这可能会让你们中的一些人产生疑问：这些光纤是如何制造的？

And that may lead some of you to the question of how are these fibers made?

我觉得这真的非常有趣。

I found this really, really interesting.

光纤的核心是光被限制的地方。

So the core of the optical fiber is where where the light is confined.

如果你光纤外部的介质光学密度更低，就可以实现全内反射。

And you can you can get total internal reflection if your media outside of the fiber is just optically less dense.

所以你完全可以只用一根玻璃丝放在空气中。

So you could have just a single strand of glass in air.

但如果你的光纤有任何弯曲或导致散射的缺陷，就会让光泄漏出来。

But if you have any sort of kinks or something that causes scattering on that fiber, that will make the fiber leak light.

因此，你通常会把核心包裹在一层光学密度稍低的包层玻璃中，以形成模式约束。

So you usually put that core in a cladding in a slightly less optically dense glass to form the the mode confinement.

但这种结构非常纤细，然后被拉成很长的光纤。

But that is incredibly thin then pulled into a very long fiber.

所以让我感到有趣的是，它的制造过程以及最初使用的材料是一种叫做预制棒的东西，这是一种更大尺寸的玻璃，其中纤芯与包层的尺寸比例是正确的，但尺寸并不相同。

So it was interesting to me how it's made and what you're starting off with is something called a preform, which is a much larger piece of glass where you get the the ratio of the core to cladding dimensions right, but not in the same size.

然后在所谓的拉丝塔中，你将它加热并拉出光纤，这样就能保留两种玻璃之间的比例，但尺寸变得非常非常小。

And then in a so called draw tower, you heat that up and you pull out the the fiber, which then retains the ratio between the two types of glass, but in a very, very small form factor.

为了更直观地理解这一点，网上有一些视频展示了人们制作糖果的过程。

So to put that into a more graspable context, there's these, these videos online where people are making candy.

就是先制作大块的硬糖，然后将它们拉成非常细小的糖果段。

So making large drawings in a in a hard candy and then pulling them out to very small pieces of candy.

光学纤维的制造正是如此，只不过方式更加纯净精密。

That's exactly how optical fibers are made, but in a much more pristine way.

因此，光学纤维本身，或者支撑光学纤维的基本原理，自19世纪中期以来就一直受到关注。

So optical fibers have or at least the the principles that underpin optical fibers have been of interest since the mid eighteen hundreds.

一些最早的实验——如果你有一个激光笔和水龙头，甚至可以在家尝试——实际上是利用水流来引导光线。

Some of the earliest experiments and one you can do at home if you have a laser pointer and a faucet, we're actually using jets of water to guide light.

如果你让水从水龙头流出，并用激光笔照射，你可能会找到一个合适的入射角，让光线耦合进入水流，然后在内部发生全反射，光线就会沿着水流的路径从水龙头中传播出来。

So if you have water flowing out of your faucet and you play around with the laser pointer, you'll probably be able to find the right angle where light can couple into it, but then it stays totally internally reflected and the light follows the path of the water out of the faucet.

一旦人们开始制造玻璃纤维，它们就被用于医疗设备，比如胃镜，很早就被用来观察人的喉咙。

Once people started making glass fibers, it was used in medical equipments, gastroscopes to peer into into the throat of people were around quite early on.

然而，以今天的标准来看，这些光纤的光传输性能非常差，仅几米长的光纤中就有大约90%的光损失。

However, the optical transmission of those fibers were were terrible by today's standards, and about 10% of the light got through in just a few meters.

因此，许多人认为光纤根本不可能用于长距离通信，因为它们最初损耗太大。

So many people discarded fiber as a way of of ever doing long distance communication given how lossy they were in the in the beginning.

然而，也有一些人持不同意见，认为光纤是可行的，其中一位著名人物是研究人员高锟，他早期就证明了如果能把玻璃中的损耗降到足够低，就能实现非常适用于通信的光传输。

However, some people disagreed and believe that they could be, and one notable person in that context is researcher named Charles Kao, who's early on showing that if you get the losses sufficiently low in glass, then you can get transmission that is is very much useful for communication.

他因在光纤通信中光传输方面的开创性成就，获得了2009年诺贝尔奖的一半奖金。

He won half of the Nobel Prize in 2009 for his groundbreaking achievements concerning the transmission of light in optical fibers for fiber communication.

阻碍光纤发展的因素之一是，在其发展过程中，我们已经能够非常高效且简便地利用电磁波（如微波）通过金属导线进行通信，这些金属导线制造起来更简单，对精度的要求也远低于光纤。

One of the things that inhibited the development of optical fibers was that during its development, we were actually able to do communications very effectively and more easily with electromagnetic waves like microwaves and that sort of thing traveling down metallic fibers, which are practically easier to make and require much less precision than optical fibers.

因此，只有当微波通信因其他根本性限制而接近其极限时——我相信如果有的话，‘电子与波’播客会解释这一点——光纤才重新成为通信领域的严肃选择。

So it was only really when the the limits of microwave communications started to be reached due to other fundamental limits that I'm sure the electrons and waves podcast will explain, if there is one, that the optical fibers became the serious intender for for communications again.

当激光在20世纪50年代首次被发明时，人们很快意识到它可以作为通信手段。

When the laser was was first invented in the nineteen fifties, people early on jumped on the possibility of using that as a mode of communication.

然而，人们很快意识到，为了使光能够传输很长距离，必须有一种方式来约束或引导光。

However, it was quickly realized that one needed some sort of confinement or way to guide the light in order to make it traverse long distances.

于是，光纤似乎成为了实现这一目标的良好候选。

And then optical fibers did seem to be a good candidate for that.

因此，在二十世纪七十年代，来自知名玻璃制造商康宁公司的三位研究人员展示了一种光纤，其在红光波段的衰减率为每公里20分贝。

So in the nineteen seventies, three researchers from Corning, which is a renowned loss maker, presented a optical fiber which had a 20 dB per kilometer attenuation or loss in the the red wavelength band.

而这一数值正是光纤传输性能达到可用门槛、足以作为通信手段的临界点。

And that was kind of where the threshold for good enough fiber transmission was in order to kick off using it as a mode of communication.

作为参考，如果衰减是每公里20分贝，这意味着光在光纤中传输一公里后，到达另一端的光功率只剩下1%。

So just for reference, if it's 20 DB per kilometer loss, what that means is that after one kilometer of transmission through the fiber, you would have 1%, the power of light that reached the other end.

这就是当光纤被视为远距离通信其他方式的可行替代方案时，所达到的损耗水平。

So that's the type of loss that was achievable when optical fibers were seen as a viable alternative to other methods for long distance communication.

是的。

Yeah.

我个人没有广泛使用分贝概念的背景，因此从未真正建立起对它的直观理解。

I I personally don't have a background where where the decibel concept has been widely used, so I've never really built an intuition for it.

对于同样处于这种情况的听众来说，分贝单位首先是对光纤中入射光功率相对于某个参考值的差异的度量。

And for the people listening in the same situation, the decibel unit is, first of all, a measure of the difference to some sort of reference value for a fiber that's the light entering the fiber.

虽然了解计算分贝损耗的公式很好，但掌握一些技巧也很有帮助。

And while it's good to know the the equation for calculating the decibel loss, I think a few tricks is is good to know.

分贝数值中的十位数代表数量级。

The tens digit in the decibel number refers to the order of magnitude.

所以当我提到20分贝的损耗时，我指的是两个数量级的损耗，也就是100倍的损耗。

So if I'm saying 20 dB loss, I'm referring to two orders of magnitude of loss that is a 100 times loss.

这与对数的运算方式一致。

And just how logarithms work out

这同时也是镓砷激光器被发明的时候。

this was also when the gallium arsenide laser was invented.

这对于光纤通信非常有用，因为镓砷材料发出的激光波长约为850纳米，处于近红外波段。

And this is really useful for optical communications because gallium arsenide emits a lasing wavelength at around 850 nanometers, which is in the near infrared.

这对于光纤通信非常有用，因为这个波长更深入红外区域，而那里通常的损耗更低。

And that's really useful for fiber optic communications because it's it's further into the infrared where losses are typically lower.

当我们研究光纤中的不同损耗机制时，如果处于远红外区域，会遇到如分子直接振动和吸收之类的现象。

So when we're looking at different loss mechanisms in fibers, if we're in the far infrared, we'll get things like direct molecular vibrations and absorptions and that sort of thing.

但随着波长越来越短，真正开始占主导地位的是所谓的瑞利散射，这是电子围绕原子振动引起的散射，是一种非常基础的现象。

But then as we go to shorter and shorter wavelengths, what actually starts to dominate is something called Rayleigh scattering, which is just the shaking of electrons around their atoms causing scattering, which is quite fundamental.

这和天空为什么是蓝色的原因是一样的，诸如此类的现象。

Same reason why the sky is blue and all that other stuff.

因此，随着波长变短，这种散射会以四次方的幅度增加。

And so as you go to shorter wavelengths, that actually increases to the power of four.

所以当波长接近蓝光极限时，瑞利散射会急剧上升。

So as you go to the the blue wavelength limit, you get this very steep increase in the Rayleigh scattering.

当波长进入极远红外区域时，其他吸收机制，如分子振动，就会开始占主导地位。

As you go to the very far IR, you'll start to get, other absorption mechanisms like vibrations that dominate.

但在这些机制中，最佳窗口正如我们后面将看到的，大约在1550纳米左右，通常比可见光激光器的波长更偏向红外区域。

But then within that, the sweet spot is really, as we'll see later, around fifteen fifty, but generally further out to the IR than visible lasers.

因此，砷化镓激光器的发明使我们能够进一步深入红外区域，当时实现了比以往使用如氦氖激光器产生的630纳米波长更低的损耗。

So the gallium arsenide, laser invention meant that we were able to push further into the IR, and at the time, got even, lower losses than were previously possible with things like six thirty nanometers like you get out of a helium neon laser.

因此，砷化镓激光器、低损耗光纤以及近红外探测器这三大概念相结合，使我们在20世纪80年代得以建立起首批商用系统。

So these two concept of the gallium arsenide laser together with low loss fibers and also near infrared detectors allowed us to form the first commercial systems in the nineteen eighties.

这些系统速率为每秒45兆比特，中继间隔约为10公里。

And those were 45 megabit per second systems with a repeater spacing of of about 10 kilometers.

为了完整起见，中继器是一种早期系统，它接收光信号，将其转换为电信号，然后增强信号强度，以防止光强衰减至无法检测的水平。

And a repeater for for the sake of completeness is a system early on, which took the optical signal, converted that to electrical signal, and then enhanced it in order to have the intensity of light not drop to nondetectable limits.

关键在于，它们能将信号转换为电信号，进行一些信号处理来清理噪声，然后再重新发射光信号，也就是驱动另一个激光器来传输经过清理的信号，这样你就不是在单纯地传播噪声了。

The key thing there is that they could convert it to electrical signal, do some signal processing on it to kind of clean it up, and then they would re emit light, you know, drive another laser to hit the the cleaned up signal so that you weren't just propagating noise, basically.

尽管这个首个商用系统的速度远不及今天的水平，但它革命性地实现了通过单根光纤同时传输约700路电话通话的能力。

And this first commercial system, while nowhere near the speeds that we have today, was revolutionary with its ability to transmit around 700 phone calls on a single strand of fiber.

因此，这是第一代真正意义上的现代光纤通信系统。

And so this is the first generation of really modern optical fiber communications.

稍后，我们进入了光纤通信的第二代。

A bit later, we entered the second generation of optical fiber communications.

在这一阶段，主要的变化有三点：一是发明了工作在1.3微米波长的磷化铟激光器，将我们进一步推向近红外区域，从而因瑞利散射而获得更低的损耗；二是我们转向了单模光纤。

And the main thing that changed as we went through this regime was one, the indium phosphide laser was invented at 1.3 microns wavelength, so pushing us further into the near infrared, lower losses due to due to Rayleigh scattering, as well as we transitioned to single mode fibers.

以前，制造足够均匀以实现低损耗单模传输且具有成本效益的光纤非常困难。

Previously, it had been hard to make fibers that were sufficiently uniform that a single mode could propagate with low loss and be cost effective.

因此，它们的尺寸略大一些。

So they were slightly larger.

它们是多模光纤，我们不会深入所有细节，但这种结构对有效通信距离有负面影响。

They were multimode fibers, which we won't get into all the nitty gritty details of, but basically that has negative consequences for the effective communication range.

将工作波长推入近红外范围使我们能够利用二氧化硅的低损耗特性制造出损耗极低的光纤。

So pushing it into these near IR wavelengths allowed us to build fibers with really low losses, exploiting the the low loss of of silica.

另一个好处是，与可见光相比，这个波段的材料色散非常低。

Another benefit was that the material dispersion is beautifully low in this wavelength band compared to the visible.

这种色散低到实际上令人惊讶。

It's it's actually quite wild how low the dispersion is.

对于二氧化硅在近红外波段，材料色散是怎样的。

The material dispersion is for NIR for silica.

如果我们计算V值，这相当于可见光波段的阿贝数，实际上会得到一个220的数值。

If we calculate the the v number, which is equivalent to the Abbe number in the visible wavelength range, you actually get a number of two twenty.

作为对比，在可见光范围内，低色散玻璃在伦琴透镜设计中非常受欢迎，以获得低色差。

And for comparison, really low dispersion glasses in the visible regime is highly sought after for Lenin lens design to get low chromatic aberration.

而可见光范围内最高的AVI数值大约为85，出现在某些萤石和磷化物掺杂的玻璃中。

And the absolute highest AVI numbers around are about 85 for some fluorite and phosphide doped glasses.

因此，近红外区域的材料色散确实非常低。

So just the material dispersion in NIR is beautifully low.

你希望低色散的原因，换句话说，为什么

And the reason you want low dispersion, in other words, why

你

you

希望所有波长的折射率都相同。

want the refractive index to be the same at all wavelengths.

有几个原因，等我们讲到第三代和第四代时，会进一步探讨这些。

There's a couple reasons and later when we get into the third and fourth generation, we'll explore more of these.

但实际上，如果你发送一个光脉冲，如果色散很高，这个光脉冲会在时间上被拉长。

But what actually happens is that if you send a pulse of light, if there's high dispersion, that pulse of light will get stretched out in time.

这实际上构成了限制，因为脉冲是由许多不同波长的成分组成的。

So that actually limits because it's composed because it's a pulse, it's composed of many different wavelength components.

因此，如果折射率不同，就意味着有些成分传播得比其他成分慢。

So if the refractive index is different, that means that some are traveling slower than others.

于是脉冲在时间上被拉宽，这开始从根本上限制你传输开/关信号的速度。

And so the pulse shake gets spread out in time that actually starts to fundamentally limit how quickly you can transmit on and off signals.

所以丹尼尔刚才谈到的玻璃在1300纳米处的低色散，指的就是材料色散。

So what Daniel was just talking about, about the low dispersion of glass at 1,300 nanometers, that's the material dispersion.

因此，任何玻璃、任何材料的折射率都会随波长或频率而有所变化。

So any glass, any material has some variation in refractive index as a function of wavelength or frequency.

通常情况下，这种色散的表现形式是随着波长越来越短，几乎呈准指数增长。

And typically almost always the way this dispersion looks is something that increases quasi exponentially as you get to shorter and shorter wavelengths.

尤其是当你超过蓝光波段时，它开始急剧上升。

So especially when you get past the blue, it starts to go up really quick.

在短波端你会得到很高的折射率，而随着波长变长，折射率会逐渐减小，并最终趋于某个长波长极限。

You get high refractive indices, and then as you go to longer wavelengths, it gets smaller and smaller and kind of settles on some long wavelength limit.

所以这是色散的一个贡献因素。

So that's one contribution to dispersion.

我们提到色散是不利的，因为它会拉伸脉冲，而这对快速连续发送大量脉冲不利。

We mentioned dispersion is bad because it stretches out your pulse and that's bad for sending lots of pulses in quick succession.

但还有另一个因素对总色散有贡献。

But there's another contribution to the total dispersion.

除了材料特性外，当你使用光纤或任何波导时，该光纤的有效折射率不仅取决于材料特性，还取决于光纤的几何结构、波导结构以及波长。

So in addition to the material properties, when you have this optical fiber or any waveguide, the effective index of that fiber depends on the material properties, but also depends on the geometry of the fiber, the waveguide, as well as the wavelength.

因此，即使材料本身完全无色散，仅凭几何结构和材料组成也会对色散产生影响，因为对于固定的几何结构，不同波长的传播波的电磁模分布是不同的。

And so just the geometry and the material composition, even if the materials were a 100% dispersionless, will have its own contribution to dispersion because the actual electromagnetic mode profile of the propagating wave is different with different wavelengths for the same fixed geometry.

因此，实际上存在两种机制共同贡献于传播波所经历的总色散。

And so there's actually two mechanisms that contribute to the dispersion of the, the net dispersion that the propagating, wave sees.

这很完美，因为对于1300纳米波长，波导色散和材料色散恰好相互抵消，从而形成了零色散点，这使得它成为第二代光纤通信系统理想的中心工作点。第二代光纤于1987年首次商用，比特率为1.7吉比特，中继间隔约为50公里。

So that's that's beautiful because then the waveguide dispersion and the material dispersion for this 1,300 nanometer wavelength actually cancel out, allowing one to have this zero dispersion point, which made it a great position to to build a communication system around in the second generation of fibers, which were first commercially available in the in 1987 with a bit rate of 1.7 gigabits with a repeater spacing of about 50 kilometers.

因此，比特率比第一代高出约40倍，中继间隔也延长了约五倍。

So about 40 times higher bit rate than the first generation and about five times longer repeater spacing.

因此，这标志着第二代光纤的终结。

So that kind of caps off the the second generation of optical fibers.

当我们进入第三代光纤时——我在上世纪九十年代初就提出了这一概念——趋势仍然是进一步向红外区延伸。

As we move to the third generation, which I've invented quite soon after in the nineteen nineties, there was still the shift towards further into the IR.

因此，我们正在远离这个完美的零色散区域。

So moving away from this perfect zero dispersion regime.

转向完美的零损耗区域。

To the perfect zero loss regime.

或者不是完全为零，而是尽可能接近零。

Or not not quite zero, but as close to zero as we could get.

因此，我们将工作波长略微向红外区延伸至约1550纳米。

So going slightly further out into the IR at about fifteen fifty nanometers.

在这两个阶段之间，你会经过一个显著的衰减峰值，即玻璃中的水峰，这是由于光纤中不可避免存在的水分子溶解所导致的损耗。

So maybe mentioning between those two steps, you go over a a really big peak of attenuation, which is the water peak in glass where you have losses caused by the inevitable solution of water molecules in the fiber as well.

这些水分子的振动模式会在1300到1550纳米之间强烈吸收光。

So these would be vibrational modes of the water molecules that absorb the light very strongly, somewhere between thirteen hundred and fifteen fifty nanometers.

因此，在1300纳米和1550纳米之间存在一个明显的跳跃，这是为了增加波长所必需的。

And that's why you're there's there's that discrete jump between thirteen hundred and fifteen fifty that's kind of necessary in order to increase the wavelength.

所以在1550纳米波长的第三代光纤中，我们已经将损耗降低到大约每公里0.2分贝。

And so the third generation at fifteen fifty nanometers, we had got down to about 0.2 DB per kilometer losses.

作为参考，这意味着经过50公里后，我们仍然保留了10%的光信号，这对于光通信来说非常出色。

So for context, that means that after 50 kilometers, we still have 10% of our light remaining, which is quite impressive for optical communications.

是的。

Yes.

把这个数字放在背景下看看。

Just put that into context.

这种损耗低得惊人，我通常觉得窗玻璃已经相当透光了。

It's it's such an incredibly low loss that I tend to think of, of window glass as quite transmissive.

我认为主要的损耗来自于光进入和离开窗玻璃时因折射率差异引起的微弱反射。

I think of the losses mainly coming from the frail reflections due to the refractive index contrast entering and exiting the window.

但普通玻璃的衰减大约为每公里1000分贝，这意味着当时最先进的光纤的衰减比普通窗玻璃低了5000倍。

But regular glass has an attenuation of about a thousand DB per kilometer, which means that the state of the art fibers were actually 5,000 times less attenuation than regular window glass.

在每公里0.2分贝的损耗下，我们实际上已经达到了石英材料所能实现的最低损耗极限。

And at 0.2 DB per kilometer, we're really at the fundamental limit, at least for silica of the lowest loss that we can possibly get.

现在你受限于瑞利散射损耗，除了更换材料或改用空心结构，否则无法消除这些损耗，而现在很多人正在这么做，因为瑞利损耗源于原子本身，而你无法摆脱原子。

Now you're, you're limited by Rayleigh losses and there's nothing you can do to get rid of those, other than change materials or go to, airspace, which lots of people are doing now, because the Rayleigh loss is caused by the atoms themselves, and you can't get rid of atoms.

真糟糕。

Dang it.

而且仍然没有数量级上的提升空间。

And it's not orders of magnitude gain still to be had.

我说过你无法消除原子，但其实你可以。

I said you can't get rid of atoms, but you actually can.

这正是现在研究人员正在做的事情。

And that's what people do in research now.

直接使用空心光纤。

Just go to aircore fibers.

这是当前光纤光学领域最令人兴奋的前沿技术。

It's the current, exciting state of the art in fiber optics.

尽管他们已经尝试了一段时间。

Although they've been trying that for a while.

正如我们提到的，零色散点原本在1300纳米，但当这个点被推到1550纳米的低损耗区域时，色散增加了，因为波导色散和材料色散在1500纳米处无法相互抵消。

And as we mentioned, the zero dispersion point was at 1,300 nanometers, but having this pushed to the really low loss regime of fifteen fifty, the dispersion went up because the waveguide dispersion and the material dispersion didn't cancel out that 1,500 nanometers.

于是，出现了色散位移光纤，也就是所谓的渐变折射率光纤。

So then there was an advent of dispersion shifted fiber, which it's something called graded index fiber.

到目前为止，所有光纤都是阶跃折射率光纤，其核心具有单一的折射率，包层则具有阶梯式差异的折射率。

So up to this point, all fibers were were step index fibers where you had a core of a single refractive index and cladding of a stepwise difference refractive index.

但你可以通过在光纤核心中引入折射率梯度来玩些花样，从而将色散移到1550纳米波长波段。

But you could play tricks with having a gradient of refractive index through the fiber core, which allowed the dispersion to be shifted to this fifteen fifty nanometer wavelength band.

所以这是两代技术。

So that's two generations.

第一代商用系统出现在20世纪90年代，比特率约为每秒10吉比特。

First commercial systems were available on the in the nineteen nineties with a bit rate of around 10 gigabits per second

中继器间距也逐渐接近100公里。

with a repeater spacing pushing towards a 100 kilometers.

然而，实现100公里的中继间距仍然非常困难，因为如果损耗为每公里0.2分贝，那么在100公里距离上就会产生20分贝的损耗。

So the push for a 100 kilometer, spacing was still quite difficult though, because at say 0.2 DB per kilometer, you're getting 20 DB loss over a 100 kilometers.

这意味着你的信号只剩下1%的强度。

So 1% transmission of your signal.

因此，要实现这一点，需要更灵敏的探测器。

So one of the things that was required in order to enable this was more sensitive detectors.

他们实现这种灵敏度大幅提升的方法是使用所谓的外差探测器。

And the way that they were able to achieve this step increase in sensitivity was by using what's known as heterodyne detectors.

外差探测器的工作原理是：你的信号由于长距离传输而衰减到初始幅度的1%，然后将其与一个参考信号结合，这个参考信号来自探测器处的一个强度高得多的参考激光器。

The way a heterodyne detector works is you have your signal that's now at 1% of its initial amplitude because of the low losses, but over a long distance, and you combine this with a reference signal, some reference laser that you have at your detector that's of a much higher amplitude.

所以你将这两者结合起来。

So you combine those two together.

就像将任意两个正弦波叠加时一样，你会产生新的频率分量。

And now similar to what you'd get when you combine any two sine waves, you get new frequency components.

你会得到由于两者调制而产生的拍频。

You get this beat frequency that's due to the modulation of the two together.

由于拍频信号同时包含信号和参考激光，其幅度高于原始信号，因此你实际上放大了信号中包含的数据，从而可以使用复杂的电子设备进行检测，并利用傅里叶变换的原理。

And so, because the beat frequency is composed of both the signal as well as the reference laser, its amplitude is higher than the signal, and so you've basically amplified the data that was contained within that signal, and so you can use fancy electronics to detect that and do the way Fourier transforms work.

你的数据信号得到了很好的保留，现在因为与参考信号进行组合和干涉而被放大了。

You have your data signal well preserved, but now amplified because you've combined it and interfered it with this reference signal.

然而，在每个中继站进行检测并将信号转换为电信号等，仍然需要在每个中继站建设大量基础设施。

However, detection at every repeater station and the conversion to electrical signal and so on is still quite a bit of infrastructure to build in each repeater station.

以及更多的新损耗通道。

And more new loss channels.

没错。

Exactly.

而且永远不可能做到完美。

And never never do it perfectly.

没错。

Exactly.

因此，人们仍然在努力寻找其他方法来增加中继站之间的距离。

So there was there was still a push for other ways of increasing the range between repeaters.

实现这一点的一种方法非常巧妙，那就是使用光放大器。

And one way of doing that, which is really, really neat is with optical amplifiers.

其中的主力是掺铒光纤放大器，它能让你将光纤纤芯用作泵浦激光器。

And, the workhorse of this was the Erbium doped fiber amplifier, which allows you to use the fiber core as a pumped laser.

因此，这是在传输窗口内对光强度进行光学增强。

So an optical based enhancement of the light intensity in the, in the transmission way window.

在这种情况下，你不再将信号转换为电信号再重新生成光信号。

So in this case, you're no longer converting it to electrical and then regenerating an optical signal.

你拥有传输数据的光纤，通过掺铒并用另一个激光器泵浦该区域。

You've got your fiber that the data is transmitting through and that with this Erbium doping, you pump that region with a, with a different laser.

当数据信道通过时，由于泵浦激光的作用，会产生粒子数反转，从而引发数据信道的受激辐射。

And as the data channel transmits through because of the pump laser, there's population inversion, which then cause stimulated emission by your data channel.

因此，你实现了信号的相干放大。

So you're getting coherent amplification of your signal.

这种掺杂光纤放大器的优点在于，其增益窗口比单波长宽得多，能够同时放大多个波长，这在后续应用中将变得极其有用。

And the nice thing about this doped Feinberg amplifiers was the fact that the gain window is actually quite wide compared to a single wavelength, which allowed to amplify many wavelength simultaneously, which will turn out to be incredibly useful for what came next.

好的。

Okay.

总结一下，我们从第一代开始，使用的是850纳米波长，当时的光纤损耗较大，但在当时已经很不错了。

To recap, we've gone from the first generation where we, used eight fifty nanometers, and we had pretty lossy but good for the day fiber optics.

我们进入了第二代。

We moved to the second generation.

我们进一步转向近红外区域的1300纳米波长。

We moved further into the near infrared 1,300 nanometers.

我们获得了损耗更低的光纤。

We got lower loss fibers.

我们相应地提高了数据速率。

We increased our data rate accordingly.

同样，第三代我们现在拥有了损耗更低的光纤，已接近理论极限。

Similarly, third generation, we've got even lower loss fibers now reaching the fundamental limit.

我们拥有了非常灵敏的探测器。

We've got very sensitive detectors.

我们在损耗方面选择了最优的波长——1550纳米，并且采用了相干放大信号的方式。

We've got the optimal wavelength choice of fifteen, fifty nanometers with respect to loss, And we've got this way of coherently amplifying our signal.

到目前为止，我们做得不错，但接下来该往哪里走呢？

So we've done pretty good so far, but where do we go from here?

好吧，你只需切换不同颜色的光。

Well, you just start blinking different colors of light.

你开始在其中使用多个信道。

You start using multiple channels within that.

这就像我有一个红色LED和一个绿色LED，同时开关它们，这就构成了两个不同的数据流。

So similar to if I had a red LED and a green LED, turning those both on and off, that's two different data streams.

我可以在1500纳米、1300纳米或任何我想要的信道上做同样的事情。

I can do the same thing around 1,500 nanometers or 1,300 or whatever channel I want.

这正是推动光纤通信第四代发展的关键，其专业术语称为波分复用。

And that's what really enabled the fourth generation of fiber optical communications, which uses this and a fancy name known as wavelength division multiplexing.

这使得数据速率得以提升至大约每秒1太比特。

And that allowed the data rates to be pushed to about a terabit per second.

1996年，第一个商用系统出现了。

In, in 1996, the first of the commercially available system was was around.

由此反思，我觉得早期的数据速率已经相当惊人了。

One reflection off of that is I think it's impressive how large the data rates were early on.

我的意思是，回想一下，2000年初我们家刚接入宽带时，速度只有56千比特。

I mean, I'm thinking back, meanwhile, in the in the early two thousands when we got broadband into my house where I grew up, that was, like, 56 kilobits.

当然，我知道那一terabit每秒是共享的，而且最后一公里的连接并不是光纤，但这也说明了技术创新与真正影响日常生活之间存在多大的滞后。

And, course, I I get that that's one terabit per second was, of course, shared, and the last mile connection was not fiber based and so on, but it kind of shows you how how big the lag between technological innovation and when it starts to truly affect everyday life is

另外，回顾一下，早在20世纪80年代初，我们第一代的速率是45兆比特每秒，而到了1996年，我们已经达到了每秒一太比特。

Also, just looking back, I mean, in early nineteen eighties that we had the first generation of 45 megabits per second, and then now, in, '96, we're at a terabyte per second.

所以我们从45兆比特提升到了一太比特。

So we've gone from 45 megabytes to one terabyte.

粗略算来，提升了七个数量级。

So seven orders of magnitude roughly.

而我们只用了大约十六年就做到了这一点。

And we've done that in, about sixteen years.

这是一条惊人的增长轨迹。

That's in a wild growth trajectory.

然后互联网泡沫破裂了。

And then the .com bubble collapsed.

太糟糕了。

So fucked up.

但光纤行业还是决定继续前进。

But, the fiber optic community decided to carry on anyway.

但公平地说，这给光纤行业带来了沉重打击。

But to be fair that that hit the fiber optic community gravely.

我的意思是，据估计，在泡沫破裂四年后，铺设的光纤中仍有85%处于闲置状态，而实际使用的容量仅占总铺设容量的约5%。

I mean, it was estimated apparently that four years after the crash, 85% of the fibers laid down was still dark, and only about 5% of the capacity that was laid out was in use.

电信泡沫附着在互联网泡沫之上，依赖于对互联网带宽持续增长的需求，推动了大量借贷用于开发新容量，而这些容量当时被认为是有用的。

The telecom bubble, which was was tacked onto the .com bubble, was kind of riding on the wave of this ever increasing capacity or need for Internet, and it spurred lots of borrowing and lending into developing new capacity, which which was perceived to be useful.

但随着互联网泡沫破裂，人们也开始对基础设施的需求失去信心，从而导致了电信泡沫的崩溃。

But then as the .com bubble crashed, the community kind of lost faith in the need for the infrastructure as well, which led to the telecom bubble crashing.

但值得庆幸的是，我们从中学到了教训，不再被炒作牵着鼻子走，作为社会，我们再也没有陷入过类似的陷阱。

But it's a good thing that we, learned our lesson from that, you know, not to get carried away with hype, and we, as a society never fell into one of those traps again.

这种事情永远不会再次发生。

This will never happen.

当然，目前并没有像AI或机器学习那样的泡沫，但我猜我们仍然处于那种……状态？

Certainly not like there's a bubble with AI or ML at the moment, but, I suppose we're still in our, what was it?

郁金香……那是荷兰的第一次郁金香危机。

The tulip what was the the Dutch the tulip first crisis.

是的。

Yeah.

对。

Yeah.

郁金香危机。

The tulip crisis.

我们仍然在重复这些错误。

We're we're still, repeating those mistakes.

我们从未从

We never learned from

是的。

Yeah.

那些错误中吸取教训，人类思维中似乎有种固有的东西让我们无法真正

From those, like There's something inherent about the the human mind that we can't really

理解这一点。

appreciate this.

尤其是当你看郁金香危机时，你会觉得人们交易郁金香期货简直荒谬至极——那些花摘下来只能撑一周，而人们却在交易根本还没长出来的郁金香，价格高得离谱。

Especially when you look at the tulip crisis, because you think it's so bloody ridiculous that people are trading like futures of fucking tulips, which last for like a week, you know, after they've been plucked and they're trading on tulips that haven't even been bought yet for astronomical prices.

然后泡沫破裂，所有人都亏得精光，失业率飙升，诸如此类。

And then it, you know, collapses and everybody loses their shorts and unemployment skyrockets and stuff like that.

这看起来太愚蠢了。

And it seems so silly.

但话说回来，确实，这比互联网泡沫或AI/ML泡沫更荒唐，但我怀疑，一百年后的人们会不会觉得，这些其实没什么本质区别。

And then, I mean, I mean, yeah, again, like it seems more silly than the dot com bubble or the AIML bubble, but I wonder if they'll still feel like it was, you know, a hundred years from now, if they'll feel that those are actually much of a muchness.

哦，这变成了一场对经济体系的抱怨，

Oh, this turned into a rant of the economical system,

但这不是关于加密货币和利率的播客。

And this is not the coins and, interest, podcast.

所以回到正题。

So back on track.

在第四代中，我们同时引入了多种不同波长的光，每种颜色都是一个独立的数据通道，这就是所谓的波分复用技术，即使用数百种甚至更多的不同颜色的光同时传输数据。

So in the fourth generation, we introduced lots of different wavelengths of light simultaneously, lots of different colors, each independent data channel, what's known as wavelength division multiplexing, where you're using hundreds or even more different colors of light to transmit data simultaneously.

这就引出了一个问题：我们接下来该往哪里走？

So that begs the question, where do we go from here?

现在我们有了多种颜色的光。

Now we have multiple colors.

我们拥有了最佳的传输距离和最低的色散等等。

We have the best transmission range and the lowest dispersion and so on.

因此，开发光通信系统的人们手中仍然还有不少妙招。

So there's still tricks up the sleeves of the people developing optical communist case room systems.

接下来他们提出的是相干系统，利用不同相位可以独立传输、不同幅度可以被单独识别的特性，来承载更多的信息。

And the next one they pulled out was coherent systems where you're actually using the fact that the different phases can be carried separately and that different amplitudes can be perceived separately to carry even more information.

这进一步提升了光纤通信信道本已非常高的数据传输速率。

And that's adds another multiplier factor onto the already quite vast data rates that you can use on these fiber communication channels.

所以，为了总结推动这些提升数据速率技巧背后的核心理念，我们希望拥有不同的信道或正交的信息流，使它们彼此独特并能独立解码。

So just to summarize the kind of ethos of what's driving all these different tricks to get more data rate, what we want to do is we have different channels or orthogonal, streams of information such that they're unique and be decoded individually.

最初，我们通过不同时段发送不同信号，即在时域上对这些信号进行调制。

So at the beginning, we were sending different signals through at different times, like in in the time domain, we were modulating those signals.

然后我们转向添加不同颜色，因为不同频率的正弦波彼此正交，因此每个都是独立的信息流。

Then we move to adding different colors because all sine waves are orthogonal to one another if they're at different frequencies, so they're each independent stream of information.

所以我们同时在时域上独立调制这些信号。

So we are modulating those both independently in time.

展望未来，当我们为实现更高的数据速率倍增而增加新信道时，我们仍在沿用相同的基本思路，即寻找不同的正交信道。

And looking forward, as we go to add new channels for more and more multipliers of our data rates, we're still playing the same basic thing of looking for for different orthogonal channels.

相位就是其中一种方式。

Phase is one way.

所以你可以有一个比另一个正弦波相位差90度的正弦波，这两个就是两个正交的信息流。

So you can have a sine wave that's 90 degrees out of phase with another sine wave, and those are two orthogonal streams of information.

同样，我们可以利用光的偏振，因为偏振方向正交的光信号本质上就是独立的信息流。

Similarly, we can use polarization with light because light that's orthogonally polarized to another stream of light is very inherently an orthogonal stream of information.

这些就是第五代光纤通信中采用的技巧。

And so these are the tricks that went into the fifth generation of optical fiber communications.

丹尼尔，这一代技术能达到的最高速率是多少？

And Daniel, what's the highest data rates that they can get in that generation?

实际上，这一代技术目前仍在发展中。

So it's, it's actually a generation that's kind of still ongoing.

甚至有一篇非常近期的论文展示了每秒402太比特的数据传输速率。

And there was even a a very recent paper showing 402 terabits per second of of data rate transmission.

这利用了整个光学波段，将275纳米宽的频带划分为1500个信道。

And that's using basically the entire optical band between so a 275 nanometer wideband divided into 1,500 channels.

他们用了一些非常专业的术语，比如双偏振正交幅度调制，这意味着系统每个星座点可以发送256种不同的符号，从而实现了通过单根光纤传输海量数据。

And they used some very fancy names here, but they used dual polarization quadrature amplitude modulation, which is basically that you can send 256 different symbols per constellation of the system, and that allowed you to then transmit these vast amount of data through a single fiber.

这402太比特是单芯波分复用相干系统检测中的最高成就。

And that four zero two terabit is the greatest achievement within the single core wave and division multiplexing coherent system detection.

还有偏振。

And polarization.

还有偏振检测方案。

And polarization detecting scheme.

这并没有真正回答数据速率能有多高。

That doesn't really answer how how high the data rates can go.

未来几代技术可能还有其他几招，能让我们进一步提升，但目前这已经非常接近单模光纤在物理上可实现的极限。

There might be another couple of tricks up the sleeves of of future generations, which allows us to push this further, but it is pretty darn near the the limit of what can be physically achievable in a single mode fiber.

但数据速率需求和互联网增长等问题依然存在，这是一个需求旺盛的领域，需要持续增加带宽容量。

But the problem still remains the data rate need and the Internet growth and so on still, it's a thirsty development and it it needs ever increasing bandwidth capacity.

因此，第六代光通信技术不再依赖单根光纤传输，而是进入了所谓的空间分复用领域。

So there is the sixth generation of optical communication where instead of relying on the transmission through a single fiber, you're entering a domain where you're called spatial division multiplexing.

接下来交给你了，史蒂夫。

Take it away, Steve.

在空间分复用中，我们只是在寻找新的正交信道来传输信息。

So in spatial division multiplexing, again, we're just looking for new orthogonal channels to transmit information.

在空间分复用中，最简单的方式就是将信号分配到多根彼此靠近的光纤中。

In spatial division multiplexing, or in the simplest form, you're just splitting it out into multiple fibers bundled near to one another.

因此，你可以把整个系统看作一根有效光纤，同时通过多个纤芯传输数据。

So you can call the whole thing an effective fiber, sending data through multiple cores simultaneously.

还有一些更复杂的方式，我们实际上回到了多模光纤，同时通过不同模式传输数据。

There are kind of fancier versions of this where instead we're going actually back to multimode fibers, and we're sending the data through different modes simultaneously.

每种方式都有其优缺点，主要取决于信号恢复的能力以及实际的限制。

And there's pros and cons to each largely around your ability to recover the signal as well as practical limitations.

比如，如果多个纤芯靠得太近，由于几何结构的原因，光纤在弯曲时容易导致某个纤芯断裂。

Like if you have too many cores near one another that hurts the ability for the fiber to bend without snapping one of the cores just due to geometry.

关于光通信的未来发展方向，目前存在两种不同的观点，因为采用空间分复用确实面临一些障碍——你需要部署更多新的硬件基础设施，比如铺设多模或成束光纤，这需要巨大的资本投入。

There's kind of two different camps in where the future of optical communication is heading because there's definitely some some hurdles with going to spatial division multiplexing because you're requiring more new hardware infrastructure to laying down multimode or bundled fiber cables, which is a very large capital investment.

另外补充一点，如果使用一代收发器，插拔不同光纤并更换部分硬件，在使用标准单模光纤时更容易实现；但当我们使用多芯光纤或多模光纤时，实际操作会变得更加复杂和困难，因为光纤的朝向变得至关重要，而这种调整一旦大量重复，成本就会迅速累积。

And just to jump in there, coupling to those things, if you have one generation of transceiver, plugging in a different fiber and changing some of the hardware there is easier with staying with a standard single mode fiber, but it gets more challenging and less trivial for how to actually do this when we have multi core fibers or multi mode fibers, especially just because now the orientation of the fiber matters a lot, and that's a cost that starts to add up when you need to do this lots and lots of times.

所以仍然有一派坚持使用波分复用技术，并试图将其潜力发挥到极致，以满足所需的数据速率。

So there's still a camp that stays with the wavelength division multiplexing and is intending to push the limits of that to get the data rates required.

是的。

Yeah.

不过我认为，即使你能实现五倍的提升，考虑到需求呈指数级增长，如果你只能做到十倍，而无法实现百倍的提升，那么追求这种方案的收益就非常有限。

I guess the thing is though, is that even if you can get a factor of five, like with our exponential increase in demand, if you got a factor of 10, if you didn't also have a way of getting to a factor of a 100, there's limited benefit in pursuing something.

实际上，你仍然需要寻找超越这一范围的其他方法。

And that in practice, you still need to look for tricks that go beyond that.

或许值得简单提一下。

It's probably worth just quickly mentioning.

你可以直接铺设更多光纤，但同样，当需要这种指数级增长时，成本会带来实际的限制。

You can just lay down more fibers, but again, that runs into practical limitations just due to cost when you need this exponential increase.

对。

Yeah.

而且，如果你真的要铺设新的光纤，

And I mean, if you're, if you're going to lay down fibers, new fibers,

想要它们

want them

那你还不如直接使用空间分复用。

to You might as well do spatial division multiplexing.

所以回过头来看，我们已经讨论过光纤通信的发展，从仅开关一个光信号，发展到在更优的波长范围和更优的光纤中开关一个光信号。

So looking back, I mean, we've talked about sort of the development of optical fiber communication going from blinking on and off one light to blinking on and off one light, but at a better wavelength regime and in better fibers.

然后我们开始同时开关两个光信号，实际上就是数百个甚至上千个光信号。

Then we started blinking two lights or in practice hundreds or even over a thousand lights simultaneously.

接着我们开始利用不同的偏振、不同的波长、不同的相位和不同的幅度来调制光信号，现在可能还加上了不同的角度或不同的位置。

Then we started blinking lights at different polarizations and different colors and different phases and different amplitudes, and now maybe different angles or different positions.

因此，我们在光纤通信方案上已经取得了长足的进步，但本质上我们仍然只是在开关光信号。

So we've gone a long way to develop these optical fiber communication schemes, but we're still just blinking lights.

我们仍然只是用火焰传递特洛伊陷落的消息。

We're still just signaling the fall of Troy with fires.

我们只是在用越来越好的方式来做这件事。

We're just doing it in better and better ways.

公平地说，对于这个领域的人来说，这真是一段极其惊人的旅程。

To be fair, it's been an absolutely wild ride to be on for the people in this field.

从最早商用系统的数据速率到现在，已经提升了整整一万亿倍，而这个数字本身已经是光缆刚起步时铜线传输容量的约一万倍的飞跃。

The data rates from the first commercially available systems to where we are now has increased by a factor of 10,000,000,000 times, and that was already a step increase of about 10,000 times from the capacity of the copper wires that were transmitting signals back when the optical fibers were starting out.

因此，它真正推动了全球信息传输的爆炸式增长。

So it's really enabled the explosion of information transfer through our world.

总之，我认为，当你从博士研究的背景出发，回顾这种创新进展时，很容易陷入自己当前的工作层面，只看到眼前的路，专注于自己面临的问题。

Just, wrapping up, I think it's beautiful when you're, when you're zooming out on the progress of innovation like this coming from a background of it, of doing a PhD, it's incredibly easy to get stuck on the level that you're currently working on and just seeing the road ahead of you and be focused on the problem that you're facing yourself.

但当你退后一步，看到整个领域研究的累积成果及其对世界的改变时，你会感到一种难以言喻的——

But when you're taking a step back and seeing what the aggregate of research in a field is doing and how that's changing the world makes you kind of

哑口无言？

Speechless?

我们已经几次总结了实现光通信数据速率惊人提升所采用的技术手段。

A couple times we've kind of summarized the technical tricks that were played to realize this incredible improvement in data rates for optical communications.

但同样有趣的是反思：当微波通信逐渐触及自身的基本极限时，是什么推动并促成了这种进步，从而催生了对更高速数据通信新途径的需求。

But it's also fun to reflect on, you know, what drove and enabled this progress as microwave electrical communications were reaching their own fundamental limits, that started to create a need for new avenues for faster data communication.