The Modem: Building The Internet With Sound
Вставка
- Опубліковано 11 жов 2021
- ▶ Visit brilliant.org/NewMind to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription
THE INTERNET
ARPANET was initially created to facilitate communications among government agencies and research institutions. The civilian ARPANET would eventually be migrated to a more modernized parallel network called NSFNET. Around this time, the restrictions on the commercial use of NSFNET would be lifted and with it came the emergence of the commercial internet service provider industry.
This shift to commercialization became the catalyst for a massive influx of money, technical advancement, and the proliferation of access that transitioned the early internet from the military’s technological marvel to the massive communications apparatus that infiltrates every aspect of our lives today.
BAUD RATE
The baud unit’s definition would be revised and redefined formally in 1926, to represent the number of distinct symbol changes made to a transmission medium per second.
THE FIRST MODEMS
Derived from the term modulator-demodulator, a modem converts digital data into a signal that is suitable for a transmission medium. A year later, a commercial variant of the SAGE modem would be introduced to the public as the Bell 101 Dataset.
FSK
In 1962, the underlying technology of the modem would split from that of teleprinters with the introduction of the Bell 103 dataset by AT&T. Because the Bell 103 was now fully electronic, a new modulation method was introduced that was based on audio frequency-shift keying to encode data. In frequency shift keying a specific transmitted frequency is used to denote a binary state of the transmission medium.
By the mid 1970s, the baud rate of frequency shift keying modems would be pushed even higher with the introduction of 600 baud modems that could operate at 1200baud when used in one directional communication, or half-duplex mode.
HAYES SMARTMODEM
The Smartmodem introduced a command language which allowed the computer to make control requests that included telephony commands, over the same interface used for the data connection.
The mechanics allowed the modem to switch between command mode and data mode by transmitting an escape sequence of 3 plus symbols. From this, the Hayes smart modem quickly grew in popularity during the mid 1980s, inherently making the command set used by it, the Hayes command set, the de facto standard of modem control.
QAM
As the early 1980s progressed, manufacturers started to push their modem speeds past 1200 bps. In 1984, a new form of modulation called quadrature amplitude modulation would be introduced to the market. Quadrature amplitude modulation is an extension of phase shift keying that adds additional symbol encoding density per baud unit, by overlapping amplitude levels with phase states. The first modem standard to implement quadrature amplitude modulation was ITU V. 22bis employed a variation of the modulation, known as 16-QAM to encode 16 different symbols, or 4 bits of data within each baud unit, using a combination of 3 amplitude levels and 12 phases.
TRELLIS
Trellis code modulation differs dramatically from previous modulation techniques, in that it does not transmit data directly. A state machine based algorithm is then used to encode data into a stream of possible transitions between branches of the partition set. This transition data is used to recreate all possible branch transitions in a topology that is similar to a trellis. From this, using a predetermined rule for path selection, the most likely branch transition path is chosen and used to recreate the transmitted data.
HIGH SPEED MODEMS
By 1994, baud rates would be increased to 3,429 symbols per second with up to 10 bits per symbol encoding now becoming possible. The dramatic boost in data rates created by TCM directly changed the look and feel of the growing internet.
56K
In early 1997, the modem would get one last boost in bitrate with the introduction of the first 56k dial-up modems. Pushing speeds above 33.6kps proved to be extraordinarily challenging as that process that digitized telephone audios signals for routing by telecommunications infrastructure made it very difficult for denser data transmissions to survive the digitizing process. This difficulty led modem manufacturers to abandon pushing analog-end bitrate speeds higher. Initially there were two competing standards for 56k technology, US Robotics' X2 modem and the K56Flex modem developed by Lucent Technologies and Motorola.
Both competing products began to permeate the market at the beginning of 1997, and by October nearly 50% of all ISPs within the United States supported some form of 56k technology. V.90 merged the two competing designs into an entirely new standard that would receive strong industry support.
SUPPORT NEW MIND ON PATREON
/ newmind
SOCIAL MEDIA LINKS
Instagram - / newmindchannel
▶ Check out Brilliant with this link to receive a 20% discount! brilliant.org/NewMind
Makes a five minute intro before even mentioning modems. Like a boss.
Go big or go home 😂
@@NewMind hell yeah! This channel is INCREDIBLY good! ...and (my trades don't matter, VERY well done!) 😊🤘🏻
It’s so that more ads can be packed in and they earn more money
@@deanbruckshaw3445 hahaha, earning money in 2021 with youtube ads... what are you, a comedian?
If you grew up with a modem you'll remember there's always the dial and handshake before you connect. ;)
I worked in telephony and early internet as far back as the mid 90s and I have never known the true origin of "BAUD rate" until now.
And as a bonus, not once throughout the entire thing did I hear that annoying dial-up modem screech sound. I guarantee that was a conscious decision. THANK YOU.
Uh.. I sorta missed it for the nostalgia factor (That doesn't mean that I'm not thrilled that we've gotten rid of them for good though!) :P
I was fascinated and kind of irked to learn that the deciBEL, being 1/10th of a bel was named after Alexander Graham Bell. Between the different spelling for some reason, and it always being used as decibels rather than bels since bels are unwieldy, it never occurred to me. But it's a shame, because there's no greater honour imo than having a commonly used (especially SI) unit named after you. So it seems like kind of an unfortunate scenario for poor Mr Bell that most people don't realize decibels are named after him. Even though most people have heard of a decibel in some context.
That was a perfect cold open. Screenwriters should take lessons from you. Stellar job, as always.
The brilliant ad? 👏 👏 prefect.
I know right? Starting with an ad is so modern and revolutionary.
In all seriousness I don't agree, it seemed very generic. Just a "jump in".
This video deserves a bigger audience. SO densely packed with high quality history and technical information. Great job! I really enjoyed it.
This whole channel is exactly that....the best!
This was good. Hope you do a follow-up on broadband.
Kinda left me hanging on how they managed to push the data rates so far with telephone lines.
Pun intended?
The presentation quality here is amazing! It's like watching that show NOVA. I was surprised you didn't mention Vint Cerf or Bob Kahn
Oh man, i miss watching NOVA now...
I think this is the best explanation of modulation I've ever seen. I appreciate that you keep it visual instead of going straight to jargon and calculus
It's because unlike most teachers, this guy actually wants to teach people stuff, rather than larping about how intelligent he is.
My parents didn't allow me internet on my own computer, so I used some coax cable (yes coax) to connect to the phone line. Resulting in my parents getting high bills.
They switched to ADSL and I wasn't allowed to have internet, so I stole their unused Wi-Fi adapter that came with the package they didn't need and it took me 2 full days to get XP to connect to the WPA2 PSK acces point.
My parents just gave in and just let me do their problem solving because they where old people that didn't understand computers. 🤣🤣🤣
I was 14, fun times, fun memory.
Very well put together! Being a DSP engineer I wrote code for several demodulators using schemes mentioned in the video (fsk, psk, qam) and I can confirm this is no easy topic for general public but this video did it really well. Greetings from Slovakia!
I work for a ISP on a hybrid fiber/coax plant.
I was so glad to see a video from this channel that I not only innately understand, but work with every single day.
I feel ashamed that I'd never heard of Ungerboeck before this video. Should be known much more widely, thanks for bringing his work to our attention!
People still rarely understand what makes TCM tick, because Ungerboek's explanation of the technique is truly opaque. It relies on the set partitionings mentioned in the video, doubling the coding space, dropping symbols, and using a convolutional code to recover the loss.
What all that machinery *really* amounts to, though, is to a marriage of coding and modulation in a manner which better approximates how analogue signals work in Shannon's continuous theory - two disciplines which are thought to be orthogonal till then. Technically, it constitutes a method of approximating the Euclidean distance (i.e. S/N ratio) between the modulated, analogue signals by the digital Hamming distance (i.e. number of bit flips in the now coded word), which then leads to a kind of in-band spread spectrum modulation which nevertheless is systematically and optimally decodable wholly in the digital domain, using the (even then well-known) Viterbi algorithm.
None of that insight is apparent from Ungerboeck's work, but was explicated only much later. In the original paper, the point is buried under the language of set subdivision which chooses symbol modulations wrt the Euclidean metric, the doubling of the symbol space which leads to many more analogue waveforms being approximable, and then the interconnectedness of those steps which leads to sequences of such symbols being easily decodable by Viterbi even after you sparsify the coding trellis once again.
Nowadays, many different high efficiency coding methods draw on the same insight, either implicitly or explicitly. E.g. OFDM is an implicit transform method in that vein, while constellation shaping is a more explicit one.
After you compress computer science in 4 chapter, now you tackle a one year worth of telecommunication engineering...
Your information density and your signal-to-noise ratio are astounding.
Wow. I am actually not disappointed.
I did not expect to see those constellation diagrams, and they were explained in a really clear way.
Keep up the good work.
Another thing to increase the phone modem throughput was echo compensation, for near and far echo.
A phone line is just two wires with approximately 600 Ohm impedance. If you excite this transmission line with a signal you want to transmit, you'll be getting this signal also as if it were transmitted by your peer.
Early modems didn't deal with this crosstalk problem and simply used different frequencies for uplink and downlink directions.
But it's possible to separate the transmitted and received signal by means of a duplex circuit, which, in its simplest form, consists of a few resistors. Such circuit is present in most POTS voice phones. The separation, though, is not ideal.
Also, because of impedance mismatch, some signal you transmit reflects on the far side and goes back to your receiver, as if it was transmitted by your peer.
Thus, V.34 modems employed digital echo compensation. During the calibration phase, they measured the local (near) and far reflection, and subtracted it by DSP code. Far echo compensation was required to allow for two hops over geostationary satellites.
Modulate, demodulate, repeat.
What a life
This was an amazing watch! I've been fascinated by modems for a while and this was a great documentary on their history and use
Idk how I missed this, all your stuff is great and somehow I missed 5 months of content.. you present wonderfully
Fully thing is, from what I can tell, that 300Hz - 4KHz limit for phone systems still seems to often linger around today, where i wish it was better. Hearing above 4KHz would definitely make calls sound clearer, especially on those with higher pitched voices. Otherwise some just sound boomy and it's just frustrating how poor call sound quality is compared to what it could be.
Nifty AF ! I find the history/evolution of technology almost as interesting as the tech itself.
Fantastic video. I even listened to it at 1x speed :)
i'm old enough to remember the transition from no internet to internet. When I was in grade school we got to 'tour' the server room for i think was gopher but could be misteaken.
how have times changed.
keep up the amazing videos and thanks to all involved, you do great work
This brought back memories, very well done.
This really brought back memories :D
As always professionally done. Thanks NM
Fun trip down memory lane. I had a bunch of the modems shown.
Shared this with my friends and liked. Fantastic stuff
Totally reminded me of that modem noise. People forget how much of the internet was built in the 90s
Oh you got me good in the beginning! I didn't even notice that you hadn't even said "modem" once until the "into" ended. Because that wasn't an intro! That was a slight of words trick, and I took it hook, line and sinker! BRAVO! :D
Amazing job! Man that kept me hooked the whole time. I wanted more!
Great video, wish UA-cam would recommend your channel to more people
Excellent video! Next video, perhaps you could cover the dsl vs. Cable "modem" wars.
Hell yes! Loved every minute of this, your content is amazing
I've found a new favorite channel.
What a phenomenal video! Thank you thank you thank you!
Fascinating stuff! Thank you.
That was bloody fantastic.
Is it ok that I'm late to work to watch this.
Guess not, but I enjoy it.
Damn you educational lust.
Superb video. It's amazing how far we've progressed in such a relatively short period of time.
Also very interesting how far back electronic data transmission goes...
*beeeeeeeep* *beep* *beep* *woorrrttttt* *dooot* *dooot* *beeeeep*
Wow, the informations in this video are so densely packed I actually had to rewind it constantly to fully understand and really follow chain of thought - I LOVE it! :D I can only imagine how much work has gone into making it - astonishingly good job ;)
Information is a mass noun and has no plural, like water, luggage, or happiness. Information in this video is densely packed. Peace. Please have nice day.
Great video! You should do this but for cable modems as well.
I've really got to congratulate you on doing an awesome job of summarising basically a full university course in one video - most of the topics here were covered in my wireless communications course (in much greater detail, of course), and this would be a fantastic introduction to any similar university course. Really excellent stuff.
+1
+1
Thank you very much!
You have disadvantaged generations of future viewers by not playing a clip of two modems negotiating the speed they would communicate at. I also thought faximile would be discussed a bit more as it was certainly a major application of the technology before computers directly. Most pc modems were 'fax' devices that were adapted. Otherwise, as someone who lived much of this and bought all the newest modems as they ve out, and implemented them at work (all the wau through VoIP) this was well done.
Man, this is nostalgia.
I remembered how EXPENSIVE the tariff was, and how SLOW it was.
1 hour of 56 kbps dial-up was like IDR 10k, while today the same IDR 10k can get me 4 GB on 4G LTE speed.
@13:26 I had flashbacks of my C=64 with a Hayes 2400 war-dialing groups of numbers, looking for unlisted connection points. Lots of fax machines and random access points. One time, I could turn on and off gas pumps in portland somewhere, but we were never able to locate the location. EDIT: I started on a 300baud modem that would usually max out at 100 baud in the beginning. I had an adapter that would allow the Hayes to connect to the C=64
Trellis modulation solves a problem of inter-symbol interference. Since a single analog symbol cannot be transmitted without affecting the symbol nearby, the readings of adjacent symbols are offset from the ideal expected positions, thus eating into the noise margin. Trellis modulation compensates for those offsets by essentially employing pre- and post-compensation, through DSP means.
In the end, Trellis wins the BER (bit error rate) war, by encoding and decoding more bits than possible or necessary, and getting more correct bits out of that.
Another approach for better usage of bandwidth and SNR was employed by Zyxel proprietary 16800 and 19200 bps modems. Those modems partitioned the frequency range into a number of bands, and then the data stream was encoded as several parallel QAM streams. Each QAM stream could use different number of levels, depending on the quality of signal in that band.
Later, ADSL used a similar approach, by encoding a block of data through Fast Fourier Transform into a sequence, equivalent to transmission on multiple frequency channels. Each channel could employ different number of bits per block. Thus, extreme non-uniformity of frequency response of a POTS line over ADSL range would not be a problem anymore.
Last time i was this early US Robotics modem was the coolest gadget ever
I made it all the way to the end of the video, and boy - the real-time footage string made me feel really uncomfortable..
Thanks for sharing. I hope to never have an encounter with a moving blade.
I remember 14.4k, 28.8k, and 33.6k and finally 56k. Was all painful until high speed cable arrived
Thank you for the info-dense videos. May I ask if every video is a summary of a book? It seems like a lot of research to get this information together. If it is indeed based off a certain book, could you recommend that?
I never could figure out how the hi-res global annihilation graphics were transmitted through an acoustic coupler in that movie "War Games"... :D
oh man and there i was at 4 years old in 85 playing manic miner on my spectrum zx.....
sweeeet!! new video!! 🤘🏻
At 15:06 it says QAM was introduced in 1984. I was working on Codex 96V29 modems that used QAM in June 1982.
While perhaps interesting, the close-up video of antiquated equipment needs a little touch up.
Where did you get find the hardware? The landfill?
Good video.
I had several 56K modems for quite a few years before I had access to broadband, and was never able to achieve 56K connectivity. It was almost always 28.8K, and occasionally I would see 33K, but that was rare.
Amazing
Of course I'm upvoting though. Of course.
Didn't dialup providers offer some sort of compression technology to up the speeds even further (for compressible data)? I think they advertised 5X speed increases. I don't know a lot about it because I had moved on to broadband at that point but I have a vague recollection of it.
Esta bien chingon te quedo poca madre te amo
ATDT ... Now there is a command that I have not remembered for many years
Basically my 2nd university year in electrical engineering (communications module) summed up in 25 minutes
No dial up screech? Ever so slightly disappointed. :D
Getting 56K out of a 56k modem is like finding a unicorn, not happening. I remember redialling to get a better connection. The best I ever got was 48K, at worse it fell back to 33.6K. After getting DSL a lot of proverbial coffee stopped being made.
Huh, I very consistently got 56k. On networks in New Zealand, too, so not some really tech heavy country.
@@mduckernz What are the odds? NZ also. We were likely unlucky. Our DSL connection was well above average, which was likely due to them installing a cabinet at the end of the street and not using the free modem. Now on rock solid fibre with my own router. They seriously give out the lowest grade garbage.
Compared to the US for day to day we are pretty slick among other things, a lot of the US is still stuck with DSL or cable with no real pressure to get better. Oz has probably overtaken the US despite the NBN farce.
I like it
I remember accidently hacking the internet in 1995. We just got a computer and internet, being very poor that was something. I never understood what is was then but i remember getting this blue screen with FBI warning. It froze our computer and my step dad had to call the number and get computer unlocked. Has anyone else had this experience or could clarify what could have actually happened?
I've never understood how binary based information (I mean 0 and 1) can be transfered via waves as see at 9:30.
Those illustrations makes me feel like I am missing something. Waves looks likes they never change so where is the difference between a 1 and a 0 ?
Is this difference created by slightly changing the frequency ? Or the power ? Or by emitting and stop emitting x times a second ? If yes how many times and how do you make the difference between "no information emitted" and "information maybe emitted but lost before the recieving end" ? Can someone explain it to me or give me a link ?
Sorry for my bad english
Bro, we needed a dial up “screeech bidee bidee, shhhhh SHHHHH” at least once. For even a moment lol
19:54 Behringer X32 paramteric EQ screen?
I came here for the sweet song of 14.4Kbps. Didn't even get to hear 300. Still great video though.
Please make a video explaining pixel led controller and software
I was expecting long forgotten modem sounds somewhere in video... :(
Funny how you used to get internet with a phone line, and now you get a phone line with the internet.
Could you please cite your sources in the video description?
@1:52 aws has been around for long time.
Watching scanned porn jpegs with 2400bps modem increased stamina and patience
bring it up to date?
What about ADSL and ADSL 2+?
Hyperlinks could have been so much more
COMPUSERVE, AOL, ROADRUNNER....AOHELL, CITADEL, GODANTI, ANTPUNT, MAGENTA
Waiting for your new video, any ETA?
After i made AssisstenKU i made Computer AssistenKU.
The fax (facsimile) machine was invented in 1843
This makes me think we actually need to split up the internet !! Like it used to!! Ban politics and entertainment rubbish!! 👍😁 Let them have there own www. 😉
"dada"
Brilliant in not the sponsor, you are!
Why is the audio in this video so quiet? Right click on the video and select 'Stats for Nerds'. From there, you can see that UA-cam measures the audio as more than 11 dB too quiet. That's way too quiet. You need to check your levels before/after uploading to UA-cam.
Peak internet was dial-up BBS porn.
Pioneering minds: We need to develops a system that allows us to communicate in the event of an apocalypse
Common every man: So we can watch porn on this right?
92% Complete, after ten minutes of downloading a short, pixelated porno clip when mum picks up the receiver at the other end of the house...
Nooo... 🤯🤬
yeah lost me at 8:19, what am i even looking at lol, baud? lol what
It's not quite correct to describe 16-QAM modulation in terms of amplitude levels and phase shifts.
More proper is to describe it as a sum of sine and cosine components, each having 4 levels.
And here 30 years later after billions of dollars have been spent, the internet is used mostly for pron.
I´d say it´s split roughly 50/50 between PR0N and cat videos. 😅
Don’t be Dramatically ..
First!
ZzzzZzzzZzzz bruh didn't even include the trademark sound. 😑
I did not hear a dial up tone once. Disappointing
I don't know why but it's difficult to understand your narration. Your speaking cadence just doesn't work.