Hacker News Reader

Comment by superkuh | Hacker News Reader

His #1 can not and will not ever happen. The radio spectrum is a shared resource. The total information capacity of the usable spectrum, say from 100 KHz to 100 GHz, is massive but most of it has terrible propagation and all of it can only be used once at a time. Massive MIMO helps in dense city cores with lots of independent paths reflecting everywhere but it's still just one spectrum in practice.

Whereas with physical transmission lines, be they cables, fiber optics, or whatever, each run can re-use the entire spectrum.

View on HN

I'm surprised his #1 is still TBD. To me, ethernet lines have become more important in the last 5 years as competitive gaming/esports has completely taken off. Latency is a far more prevalent in gamers minds today, I'd argue moreso than bandwidth and the first networking related advice a gamer receives to make sure you are on ethernet.

I see frustrated WiFi users all the time, but I've never seen a frustrated Ethernet user.

If I had a dime for every WiFi user in a video call who's "breaking up", I had a whole bunch of dimes.

Exactly. I have a suspicion that a large part of people complaining about "Internet issues" are faced with WiFi issues. Apparently many struggle with the concept of separate links that make up a connection (LAN vs WAN).

But the reverse mix-up also happens: Hotels, restaurants and the similar businesses like to boast their "included free WiFi" when they really mean complimentary Internet access, provided with an AP.

I'd take that as a given for 99% of cases. As fibre takes over from piggybacking off copper, IME actual internet issues seem to be pretty rare: [UK and] I haven't noticed that kind of service issue for a few years now, whereas used to be relatively common.

Happy to keep the pretence up though -- what else is a as neutral and widely understood as "internet issues"? Whoops, seems as if my connection has dropped out! Yes, I was very interested in that discussion over sales targets that's been going on for the past half an hour, please email me a summary! Bloody internet, always dropping out!

I always laugh when I see Comcast commercials for "fastest in-home Wifi". An 802.11ac router isn't going to help anything when you can only get 10/10 bandwidth on a good day.

I love cables while I'm using them. I hate cables while I'm cleaning. It seems almost impossible to have a clean cable setup that doesn't become a rats nest behind the desk.

Velcro cable ties can help.

My cable management is based on velcro strips nailed to the desk's edges and undersides in order to keep cables off the ground. It keeps the cables off the ground (and also mostly off the desk surface) and looks reasonably clean from eye level: https://imgur.com/a/tRamJ65

I use a set of G.hn ethernet over powerline adapters in my (small, rented 2 bed) flat. I can get 200 Mbps with no latency, packet loss or drop outs from the modem in the living room to the office upstairs.

This is a distance of about 10 meters with 2 brick Victorian internal walls in-between. No matter how much you spend on WiFi equipment you can't get more than 20 Mbps with high packet loss here.

Even downstairs, with no walls to worry about, only 5 Ghz is usable. 2.4 GHz is completely occupied by neighbours, and the lower 5 Ghz channels are all crowded out because of compatibility too.

I hate WiFi with a passion.

Mind sharing which adapters you are using? WiFi works for my situation at the moment but would be useful to know in case I need an upgrade.

I have four adapters in my house, I have one TL-WPA7510 and three TL-PA7010. They are sold as pairs so when I wanted the version with an access point built-in, I had to buy another wired version. But, you can add them individually to an already existing network. The Homeplug protocol is a standard so as long as you buy the right versions, you get max performance. These things are great. I have one in the basement for the Verizon router to plug into, one on the second floor for my PS4, the AP version behind the 4K TV in the family room, and another at a desk in the guest room. The Verizon router has a lot of interference nearby so quality is poor in some spots in the house. Quality is good, it sometimes reaches gigabit speeds but it's way more robust than wifi extenders. The apartment I lived in previously was built like a brick shithouse so wifi dropped off pretty fast once you got to the other end.

These: https://www.devolo.co.uk/magic-2-wifi-next

I hate high rate data links over power lines. Power lines are not impedance controlled. At every bend, every approach to some metal in the building, they are going to radiate interference and accept/receive interference from their environment. Using powerline networking is irresponsible and rude. The fact that any are approved by the FCC at all is entirely due to contrived testing setups that are never replicated in real building wiring.

If you use real transmission line like ethernet cable (~70 ohm twisted pairs) or coax the impedance remains constant and they might even have a bit of shielding. You'll get faster, more reliable speeds and pollute the radio spectrum significantly less.

Sometimes you can't do that if you don't own the tenement you're living in.

That doesn't prevent you from running cable along the walls and under/over doors.

- It's fiddly and ugly compared to hidden cables. I wouldn't do that in my own home, I'd run them in the walls.

- It's disruptive and time consuming. It would likely take me a whole day

- I don't see why I should invest further in improving the property when I will likely move out in 1-2 years time and won't reap the long-term benefits. I've done that already in other areas.

- It's hassle. My landlord could protest that I've done a shitty job when I move out, or, even if I haven't, make me remove it.

- I need multiple WiFi adapters anyway to cover both upstairs and downstairs effectively.

- It's not as flexible (in terms of moving things around) as plugging an adapter in to any electrical outlet.

- I needed a quick solution when I moved in so I could work.

Because ethically you're almost certainly negatively affecting the people around you and legally you're almost certainly violating part 15.

You're both breaking the law and being an asshole for aesthetics and your convenience. That's why.

I'm not in the US.

Your country almost certainly has rules about emissions in the HF frequency ranges too. Even milliwatts of radiated interference at these low frequencies will bounce around the world and interfere with everyone.

Video calls too. I just got back from visiting my in laws since everyone is now vaccinated and one of my projects while I was there was running ethernet lines to their work areas. They're both on daily video calls now and poor wifi performance was driving them crazy.

> since everyone is now vaccinated

You don't need to justify yourself. I see this more and more and it's honestly just annoying.

Beam forming, phased array antennas, sophisticated coding (CDMA) and "time slots" (TDMA) will provide a lot more availability than purely looking at bandwidth available.

I still agree wired/optical is best for most fixed installations both LAN and WAN, but people are getting more out of wireless than I would have predicted. And the "last mile" capacity of today's technology far exceeds what people seem to want even looking forward a decade...which paradoxically suggests that wireless might be adequate in the interim for some use cases.

None of these technologies allow you to exceed the available bandwidth. They just make use of the shared bandwidth more efficient. It's still a shared medium.

Where I live, a lot of people have 3G internet because mobile data is pretty cheap and the companies advertise it as an alternative to cable. And now they all have really crappy internet. In the evening when everyone watches youtube you get a fraction of the advertised bandwidth.

With fibre, every customer gets the full spectrum. And since the frequency of light is a lot higher than radio frequency, you also get a lot more bandwidth. At radio frequency we're already hitting the physical bandwidth limits; with optical transmission there's still a lot of bandwidth left.

Thinking that radio frequency transmissions are an alternative to cable / fibre is pretty short sighted thinking. Data usage is going to grow, more devices are going to use data, and wireless transmission is going to seriously limit us.

You are dead on. We are still transitioning from HD to 4k with 8k coming eventually. The difference in HD to 8k is 17x the bandwidth.

>The difference in HD to 8k is 17x the bandwidth.

The pixel is 17x, not bandwidth. Even Compressed RAW size dont scale linearly with pixel count. I dont have any experience with 8K, but compressing / encoding 4K with HEVC or AV1 tends to easier with fixed VMAF score than a comparatively low pixel count of 2K / 1080P. I would imagine the same if not better for 8K. And that is discounting the use of much better video codec like VVC which brings another 40 to 50% reduction in bitrate.

Good point.

Unless you're displaying on movie theatre sized screens, 8K seems like a waste of space/bandwidth. Even 4K is generally overkill for the typical living room.

I think we're hitting the point with video resolution that music CDs hit with audio, where improvements in fidelity are largely outside the range of human perception. It's one of the reasons the music DVDs and SACDs never really caught on.

1080p at 65 34ppi

4k at 65" 65ppi

8k at 65 135ppi

This is well within what someone with good vision can see at for example 6-8 feet

For a personal reference I could tell the difference in clarity at 8 ft between a 1080p 24" monitor and a 28"4k monitor. That is 157 vs 92 ppi on a screen a fraction of the size 5 minutes ago.

I must imagine people making such claims have poor eyesight or are using optimum viewing charts as a proxy for distances wherein human vision was sufficiently acute instead of looking for themselves.

For reference someone with good vision ought to be able to distinguish up to about 115ppi at 8ft according to this source.

https://www.quora.com/What-is-the-highest-resolution-humans-...

The source seems valid despite the mixed bag that is quora. Especially since I and many others can verify this in 10 seconds.

Edit: I wrote 43 instead of 34 by accident. 65" and 1080p = 34ppi

> Especially since I and many others can verify this in 10 seconds.

Try it with a high resolution video you've transcoded locally (so you're not dealing with streaming quality differences) at various resolutions. The human visual system is a lot less precise with video than it is with still images. A paused DVD frame often looks like a blocky, terrible mess, but it's perfectly fine when displayed as video.

I won't argue that if you pick your content carefully, and know exactly what to look for, you can tell the difference. But, practically speaking, most people watching a decently encoded video won't see a huge difference from 1080p to 4k at typical screen sizes and viewing distances, and will see even less difference going to 8k.

I honestly can't tell you if I'm watching 720p or 1080p content on our TV without looking at the source file info.

You're confusing "What can be perceived" with "What actually matters." I'd argue 4k was a bit of a stretch to sell new TVs, and 8k is going to be even more so.

We've hit, as someone else said, the 44kHz/16bit point of video. Encoding quality and bitrate matter a bit, just like mastering does with audio, but we're past the point of diminishing returns at this point.

If you can't tell the difference between 34ppi and 65ppi and typical viewing distances your eyes just don't work very well and you just literally don't realize most people see better than you even if they need glasses to do so.

We aren't talking about near the limits of human perception we are capable of perceiving 115ppi at a typical 8 ft viewing distance. 8K for a 65" screen is about the limit. 4k is itself about 1/4 the resolution we can perceive and 1080p is 1/16. You are saying that people can't on average distinguish between 1/4 the resolution we can perceive and 1/16th.

The link I posted was the informed opinion of an electrical engineer that worked in display tech. If people really can't tell the difference between 4k and 1080p shouldn't there be scientific research by this point to show this? You can't because there isn't.

You are like a color blind person exclaiming there is no difference between green and red! There is.

I agree with Syonyk here. And since you bring up quality of vision, I'll note that I do have 20/20 vision because I'm wearing prescription glasses.

> You are like a color blind person exclaiming there is no difference between green and red! There is.

You're arguing against a complete strawman here. The key point that Syonyk was making is this:

> I won't argue that if you pick your content carefully, and know exactly what to look for, you can tell the difference.

If you're watching a nature documentary or anything else that puts an emphasis on the visuals, people are going to notice the difference by themselves. But most of the stuff we watch doesn't focus on visuals, it focuses on narrative. If I'm watching a legal drama, I don't really care if I can count the hairs in the lawyer's nostril or not. Again, I'll probably be able to tell apart the 4K version from the 1080p version from the 720p version of the lawyer's face's closeup, but even if I watch just the 720p version, it won't significantly impact my enjoyment of the narrative.

Although its not one singular person arguing I can't help but feel like the goal posts are pretty mobile here.

First goalpost high dpi video is already outside the range of human perception.

> I think we're hitting the point with video resolution that music CDs hit with audio, where improvements in fidelity are largely outside the range of human perception.

Second goalpost ok its not but its really close

> I won't argue that if you pick your content carefully, and know exactly what to look for, you can tell the difference. But, practically speaking, most people watching a decently encoded video won't see a huge difference from 1080p to 4k at typical screen sizes and viewing distances, and will see even less difference going to 8k.

Third goal post ok its actually clearly discernible but I mostly care about the story anyway.

> But most of the stuff we watch doesn't focus on visuals, it focuses on narrative. If I'm watching a legal drama, I don't really care if I can count the hairs in the lawyer's nostril or not.

Believe it or not but some people said the same about HD vs SD towards the beginning. It was complete bullshit then too. I too enjoy the story and psychological drama but its not books on tape. Film and TV are a visual medium wherein being able to see the world more clearly adds to the experience.

Well of course presenting three seperate people's ideas, some of whom are refuting each other, as one continuous argument will come out looking ridiculous.

The first guy is a gentleman with bad eyesight watching a 43" tv from 16 feet away. If he watches it from any further away he wont be able to tell if its on. The second is inaccurate. The third is a complete cop out.

... wait, what? 8 foot viewing distance with a 65" screen? Ok, we have very, very different definitions of "typical" here. You're using a 50% larger TV from half the distance.

I just measured our living room. We have a 43" (4k) TV at about 16' viewing distance. Just about every chart I can find on TV resolution vs viewing distance (assuming 20/20 vision) indicates that I'm in the "720p is fine" region, on the border of "You really won't notice a difference from 480p." Which is consistent with my experience. I can see the difference between 720/1080/4k if I'm right on top of the TV, but once I'm back where I actually watch from, anything I notice is more a factor of bitrate than resolution - I can certainly tell the difference between a low bitrate MPEG2 source and a high bitrate h.264 source, but that has nothing to do with the resolution.

I don't really have the motivation tonight to pull out my trig tables (I'm about to watch a movie instead, of what resolution I honestly don't know), but... at least for my case, yes, I'm saying I can't tell the difference. And it's entirely possible my glasses need updating (quite likely, honestly), but the "TV size vs distance" charts I've found say nothing of what you're claiming either. And they're by people trying to sell TVs!

Samsung says suggested diagonal size = viewing distance /2. So at 16 ft 96".

To get a more precise fix you can break out a calculator or use this one based on THX recommended viewing angle.

https://myhometheater.homestead.com/viewingdistancecalculato...

It suggests that your tiny tv ought to be optimally viewed from 4.8 feet away and at 16 feet away you really ought to have a tv greater than 100 inches.

Of course most people don't have 100+ inch tvs but they also don't watch tv from 16 feet away unless they are watching a 45-60 FOOT theatre screen. Play with some values in the calculator if you want to see how other people use tvs.

People normally sit 8-10 feet from 60-75 inch tvs and use 40 inches for smaller rooms like bedrooms where they are 6 foot away.

To be clear about the whole picture you are a fellow with bad vision watching a small tv from nearly 3 times further back as the rest of planet earth.

At that distance your tv literally does already exceed average visual acuity and for your usage you are absolutely correct.

Meanwhile the rest of us will benefit from up to 8k in the future.

> To be clear about the whole picture you are a fellow with bad vision watching a small tv from nearly 3 times further back as the rest of planet earth.

My vision is perfectly fine, thank you very much, as you seem to insist on going on about how horrid it must be. The difference is entirely explained by the fact that I watch a reasonable sized TV from a perfectly sane distance in a living room that isn't a home theater room and has space for plenty of other activities. No idea how you get a TV 6' away in the bedroom, though, unless it's literally at the foot of the bed. I've never had one in there and never intend to.

Your linked calculator suggests that the "optimum" distance for my TV is about 6' away, which I find quite absurd (having just tried it). I'm watching movies, not programming on it.

We appear to have rather divergent views (and social group priorities, if yours has 60-70+ inch TVs at your suggested 8-10 foot viewing distance) on the nature of television viewing, and, as such, there's not an awful lot more to discuss.

I don't think its a matter of divergent priorities. The argument up thread was that we had reached the zenith of resolution based on visual acuity. What we established is that this is almost entirely true or false based on the persons vision, size of screen, and viewing distance. Based on how the majority of the public uses TVs we have provably not reached such a point.

In point of fact a medium sized living room is 12x18 with a TV on either a stand or the wall along the longer wall and seating that is around 2 feet deep and off the wall by at least a foot. This means that in an medium size living room there is 8-9 foot between seat and screen. In a large living room 15x20 there is still only 11-12 feet.

People actually are putting 60-75 inch screens therein so we can support the position that the figures given by the calculator represent how people actually use TVs and supports the position that HD -> 4k -> 8k will be perceptively still beneficial given the constraints of human eyeballs.

I'm interested in sources that establish "how the majority of the public uses TVs". I've personally never seen anyone use TVs in line with the recommendations from TV manufacturers screen size and viewing distance. Maybe this is a USA cultural thing (or, contrary, maybe my own observed viewing habits are a cultural AU/NZ thing).

What about a house size and falling TV prices thing? Half of the US couldn't sit 16 feet from their TV if they tried.

1/3 of the US lives in apartments and 80% live in urban area where housing is comparatively more expensive. Even for slightly larger houses its common to have a living room and a family room and give up a lot of space to the individual bedrooms. It's supper likely that the TV is in a room that is between 12-18 and 15-20. With the TV on the long wall.

Also for some reason sticking your TV on a stand is still pretty popular vs putting it on the wall. This means that in a 14' room your tv is between 0' and 1' off the wall and your rear end is 3' off the wall. This puts as little as 8' between viewer and tv in 12-18 and as much as 12' in the larger room if they hang it on the wall.

How the US population is distributed is kind of interesting.

https://www.washingtonpost.com/news/wonk/wp/2015/09/03/how-s...

Regarding the question of average living space in the US people in the US almost universally prefer single units detached housing but as the song goes you can't always get what you want. Many for example would prefer to be close to urban settings with more jobs. Others can only afford to rent.

Among renters, around 1/3 of the population, half live in 111 sq meters or less

1/3 of owners live in domiciles of 167 sq meters or less and and 60% of owners live in domiciles 222 sq meters or less.

Combining renters and owners 77% of Americans live in 222 sq meters, 54% live in 167 sq meters or less, 25% live in 111 sq meters or less. Some rounding and conversion between sq ft and meters.

This means about a 1/4 each live in

<= 111 sq meters

112-167 sq meters

168-222 sq meters

> 222 sq meters

https://www.frugalfringe.com/calculators/compare-your-homes-...

The people paying lots for tiny apartments in the city think your 16' of open space in your living room is as unusual as you think they are with their relatively giant TVs in smaller spaces.

I'm not the person 16' from their 43" TV ;)

Interesting, but doesn't tell me much - I live in an apartment, well under 111sqm, and have a 55" TV around 12' from the couch - so still too small / too far away, according to recommendations. People with even smaller living space (at ~80sqm, my apartment is reasonably large for a two bedroom unit where I live) may simply get smaller TVs to compensate, still being "too small".

These are anecdotes. I'm curious about what people actually do, I don't know, and (contrary to your assertions about what most people do) it doesn't sound like you do either.

I showed with data how many people live in relatively small spaces and asserted but didn't prove that Americans often devote much of their space budget to bedrooms and houses which have both a living and a family room wherein they had more budget to spare. This last shouldn't require a ton of proof.

My assertion is that due to the above asserted space constraints the majority of people in America are 8-12' from their TV. I also assert that the same people are buying 60"+ TVs

I linked a source that suggested that people could distinguish up to 115ppi at 8' If I understand correctly that means they ought to be able to distinguish about 90ppi at 12'.

If these assertions are correct most people would benefit from 90-115ppi and based on THX recommendations would benefit from the increasingly popular and cheap large screens that people are provably buying despite living in not so huge spaces. In fact tv size purchases are trending upward based on what people can afford.

https://www.consumerreports.org/lcd-led-oled-tvs/tv-trends-t...

A 60" 4k screen is only 73 ppi meaning we aren't beyond human visual acuity yet and 4k at 75" we are down to 59ppi.

The majority of consumers will based on good visual acuity be able to distinguish between 4k and 8k in the room sizes and screen sizes where such products would be relevant and applicable and a median American would benefit.

> Samsung says suggested diagonal size = viewing distance /2. So at 16 ft 96".

Now I wonder what Samsung’s motivation for this could possibly be?

You've a reasonable point but the other poster said

>but the "TV size vs distance" charts I've found say nothing of what you're claiming either. And they're by people trying to sell TVs!

The people selling TVs actually say you ought to buy a big ass TV.

Not sure about that. With large screens 55" at home and over I can see a difference between FHD and 4K, enough to know I am having network problems when the streaming service starts to downscale the resolution in a 4k video.

that could be the bitrate more than the resolution. streams are highly compressed. would be interesting to see whether you prefer 1080p bluray over a 4k netflix stream (I suspect yes).

> The difference in HD to 8k is 17x the bandwidth.

Not practically. NetFlix will just compress it down to 3Mbit anyway like they do with 4k, but be able to call it "8k content!" and (probably) charge a premium for it.

It remains to be determined.

Quadrophonic sound did not replace the "good enough" stereo even though humans can move their heads. Of course an 8K TV is just as easy to set up as a 4K one, unlike a quadrophonic stereo. And the functionality of quadrophonic records became available as "surround sound" which some people do have.

It's possible there are enough people who will appreciate the difference for 8K to become established. Personally I doubt it, but it's certainly possible.

Another possibility is that your TV watches you and provides just a 4K or 4K-ish image unless you walk close to your TV at which point it displays a higher resolution image where you are looking. Possibly with some hint-driven mix of AI upsampling and more image data.

The opposite could be true of course with 8K VR rigs in every home. I also doubt it, but it's quite possible.

The 8K-production -> 4K result is well established and will remain even if my guesses above turn out to be correct. That needs a lot of bandwidth but not at the point of viewing.

Level 3/CenturyLink/Lumen CEO reassured panicked investors and employees scared that their company would be worthless with 5G and beyond by basically saying that the last mile will increasingly become the last tens or hundreds of meters and that fiber is still the infrastructure on which these increasingly dense base stations depend on. And that the “edge computing” will likely live on the cabinets owned by the fiber provider.

It makes sense to me. Just add more fiber and let people access them over whatever.

My biggest gripe is we could choose to do away with licensing fees and spectrum auctions and open mm wave 5G to be something like Wi Fi but we are shortsighted as usual.

>Where I live, a lot of people have 3G internet.

Well, yes 3G is Shared Spectrum.

>None of these technologies allow you to exceed the available bandwidth

Exceed available bandwidth of what? Per Spectrum? Shannon–Hartley theorem?

The whole point of 4G, and 5G, mentioned in the GP as Massive MIMO was that we could workaround those limits with more Antenna. Everything we are doing today and aiming to do in 3GPP Rel 17 in a few years time are literally impossible to even infer about in the early 2000s. When Massive MIMO, or it was known as Very large Array of Antenna was first published people called the idea "crazy". And CoMP, whether the marketing decide to call it 5.5G or 5.9G along with distributed antennas being worked on in 6G.

There are no fundamental technical reason why we cant have a fully wireless Internet. Although there are many business and economical reason why this may never happen.

>>Where I live, a lot of people have 3G internet.

>Well, yes 3G is Shared Spectrum.

Sorry, I think I meant 4G, not 3G. It's marketed as LTE here.

> Exceed available bandwidth of what? Per Spectrum? Shannon–Hartley theorem?

No, not Shannon-Hartley. That's just a mathematical model.

When EM waves propagate, they are subject to diffraction, which limits both the amount of information that can be transmitted per time interval, and also the spatial resolution of the transmission. Even with antenna arrays or distributed antennas you can't get past diffraction limits; you can just get closer to them.

To get around diffraction limits, you need to use higher frequencies / shorter wavelengths. (Which has the side effect that you loose the ability of signals to go around / through obstacles, so you need a lot more cell towers.)

We're at a point where new technologies just make different trade-offs (eg. shorter wavelengths for areas with lots of wireless clients vs. longer wavelengths for sparse areas).

With a fibre connection, you don't need to make these tradeoffs; you just need to dig up the ground and you can have as much bandwidth between two points as you want.

A bit of a nitpick, but most residential fiber deployments are PONs. With a PON, a single fiber gets split into a lot of separate wavelength channels with prisms. It's still tons more usable bandwidth available than wireless.

https://en.m.wikipedia.org/wiki/Passive_optical_network

> None of these technologies allow you to exceed the available bandwidth. They just make use of the shared bandwidth more efficient. It's still a shared medium.

The beam forming and such slice bandwidth availability spatially (that's what the cell network does too) so more people can use the shared medium by...not sharing it! It's not like AM radio that gets sent in all directions regardless of whether there is someone in a given direction to listen.

> With fibre, every customer gets the full spectrum.

Well by definition that's true whether fibre or copper, but that fibre or copper is itself aggregated into connections to upstream provider, so you're just kicking the problem down the road. You don't really get the full bandwidth end to end.

As I said I think fixed wireless is an acceptable transitional technology but wired makes more sense longer term for most locations.

>The beam forming and such slice bandwidth availability spatially (that's what the cell network does too) so more people can use the shared medium by...not sharing it! It's not like AM radio that gets sent in all directions regardless of whether there is someone in a given direction to listen.

Nope. In theory yes, but unfortunately diffraction is going to get you every time if youre not transmitting in deep space.

You can add in quadrature amplitude modulation. Reusing bandwidth is a a fascinating field. It's not as simple as 1 bit per Hz

Current laser com systems are getting multiple bits per photon - it seems that the communications field will keep pushing the limits until there is simply no business case, and then push some more. Really, the only limit is Shannon's limit, but that's only concerned with data rate. The modulation schemes that sit on top of the raw data is where all the capacity magic happens.

I assume you're referring to something like this: https://www.sciencedaily.com/releases/2017/02/170203102740.h... but this isn't a scalable transmission system since it's based on single photon detection, and single-photon detection is not a high bandwidth communications ability because you have to discriminate the photons. It's a far cry from "current laser communications".

AFAIK the fastest IP network speed at the moment if 44.2 Terabits/sec for a single light source (https://www.sciencedaily.com/releases/2020/05/200522095504.h...).

I'm talking about the laser communication satellites that NASA et al have launched into orbit, which are supposedly around 3-4 bits per photon.

These are hitting upwards of 10Gbit speeds (not quite 44.2 Terabits) but still impressive nonetheless.

https://en.wikipedia.org/wiki/Laser_communication_in_space https://en.wikipedia.org/wiki/Optical_PAyload_for_Lasercomm_...

> the only limit is Shannon's limit, but that's only concerned with data rate.

But that limit, the capacity, is also the maximum possible limit of information transfer on the channel. Modulation schemes don't increase it or change it, they just push the achievable information rate of the system closer to the capacity.

Modulation is definitely the wrong word in this context. I was thinking about channel coding and compression algorithms that increase the efficiency of the system without requiring additional capacity.

This comment betrays a lack of understanding of some of these multiplexing technologies. They minimize wasted bandwidth due to signal collisions and interference, but they don't increase the overall bandwidth available.

Let's take TDMA as an example. TDMA means that instead of using the available bandwidth continuously, each participant only gets to use that bandwidth for a fraction of the available time. Saying that TDMA helps us increase the available bandwidth is like saying queueing up at the restroom will "provide a lot more availability than purely looking at the number of stalls".

CDMA is more complicated but it's still a similar story. Look at this diagram of a CDMA signal:

https://en.m.wikipedia.org/wiki/File:Generation_of_CDMA.svg

Notice how the "data" actually being transmitted is only a couple of bits, but the CDMA signal includes many more transitions. CDMA is essentially using N bits of transmission bandwidth to send a signal bit of data signal, the benefit being that if multiple signals interfere it's possible to extract one of them using some complicated math. It's like if 10 people were sharing a phone line, and instead of taking turns talking they all spoke at the same time, but repeated themselves 10 times so you could pick up enough snippets from one speaker to understand what they were saying if you concentrate hard enough.

TDMA/CDMA don't bypass information theory, though. Shannon always wins in the end.

You can improve the situation with spatial beamforming, polarization isolation, mmWave bands, etc, but there eventually comes a point where the added complexity doesn't win over just hiring a few guys with a backhoe to run fibre down the street.

Not sure why this is responding to my comment, but I was basically making that same point.

Oops, I meant to respond to the grandparent post. Sorry!

No worries!

A lot of confusion in the replies below. Here is an illustration of this point: Suppose you’ve saturated your communication bandwidth between two points, sending bits through the entire 500 Thz spectrum. If you have a wired connection, you can lay another set of cables between these points and send more bits. With wireless, there is nothing more you can do. No amount of beam forming helps you.

Intel was so pushing WiMAX at one point. I really irritated someone there when a wrote something critiquing their efforts. We are seeing wireless technologies (including satellite) starting to handle some use cases where it's hard to wire. But I do expect denser areas to remain mostly wired.

Yup. I’ve always questioned people who think wireless can replace a wire at scale. It just can’t.

That being said I can browse most sites and do 90% of my internet just fine either on my phone or tethered to my phone using my data plan. So maybe I’m wrong!

Well wireless won't ever replace wired for every use case, I don't see why we can't get to a point where 90% of households are provided internet service via cellular technologies, either to a MiFi-like device or with cellular built into the computer. If my iPhone, iPad, and Apple Watch have cellular built in, why not my Mac too?

The radio is the most trivial part possible in fact computers with cellular radios have been a thing for a very long time.

The little micro cell tower alternative arrangement exists to support devices that expect to communicate with a cell tower where cell service is poor. You normally actually plug them into your wired router so they are actually for areas where relying on wireless would be the worst possible experience.

Fiber is able to provide Gbps to for example all homes in a square mile where they each get Gbps and can indeed fairly heavily use it. Cellular internet isn't actually wireless you run fiber to the towers and then everyone in that square mile.

Urban areas where 80% of people live have a high density so for example in New York City that 1 sq mile contains 27000 people.

Wireless makes sense in the country side, because it's expensive to pull optical fiber for just a few houses.

In cities however, where most of the population is:

- Bringing optical fiber is cost effective because of the population density

- Wireless gets clogged very quickly (also because of the population density). Additionally 5G is even worse than 4G at going through walls.

Does energy cost come into play here?

Quality of Service is definitely one of the metrics ISPs track, wired is a more reliable way to guarantee service than wired in a fixed setting, upto the house.

But why would you want to have that? Simply put, with wireless you are trading the convenience of having no cables for essentially everything: Wired has better throughput, lower latency, lower bit-error rates resulting in less retransmissions (better goodput).

I'll admit there is one aspect where wireless might be interesting: Comparatively lower cost of initial set-up, especially when infrastructure is lacking. In other words, if you're trying to provide Internet access to a developing country (little existing cable/telephone infrastructure), wireless might be an interesting intermediate option. But if you're looking at a industrial nation where (rough guess) more than 90% of households already have some kind of cable/telephone/fibre infrastructure, wireless would be a downgrade.

This sounds like a very "faster horses" kind of comment.

Simply put, with SSDs you are trading the convenience of having faster boot times for essentially everything: HDDs have (a lot) more space, lower prices, longer lifespans.

Simply put, with electric cars you are trading the convenience of charging at home for essentially everything: Gas has higher range, maintenance is easier to find, filling your tank takes 2 minutes instead of hours and that tank's capacity doesn't diminish over time.

Simply put, with screens you are trading the convenience of not having to deal with paper for essentially everything: Printers have better resolution, paper is easier on your eyes, it's more portable.

I guess people just like convenience.

Wireless is a worse product than wired, but you trade the worse product for the convenience of not having the wire. None of the examples you give match that pattern.

SSDs are a better product than HDDs, they trade capacity for speed, but if you need capacity it's likely you don't really need the extra speed, and you're likely running into other bottlenecks anyway (a HDD can still saturate 1G Ethernet)

Electric cars are a "worse" product, but for most use cases they're functionally equivalent. Most people don't drive farther than the range in a day, maintenance is at the same place as ICE, slow recharge times are easily negated by just plugging in every night and charging while you sleep.

Paper is a worse product than a screen because a screen can be updated. It's about as easy to read, the resolution is about the same, they're equally portable (see a cellphone), but you can put just about anything on a screen and update it many times a second. You can't do that with paper, which is why screens have mostly replaced paper.

You connect literally all those things to a wifi hub in your house most of the time, and the likelihood of big data consumption on all of them is quite low.

It's a huge difference to think you're going to be servicing all rather then some user workstations data requirements with just todays celltowers - which isn't to say you couldn't make it feel similar, but at some point you're cramming a 5G tower into every apartment and those still need fiber backhaul.

Directional antenna completely break those bandwidth limitations. It’s not currently practical for hand held devices to make significant use of it, but ultimately everything is point to point.

As someone working on 5G, I disagree strongly. 5G supports up to 1 million devices per square kilometer (the most dense city, Manila, in the world has about 50k people per square kilometer). My understanding is that you can have about 100k devices per square kilometer before speeds drop below ideal (20 gigabit). So even if everyone in downtown Manila wanted stream seven movies at once they could all do it. As other commenters have said, that’s because of clever techniques like beamforming and OFDMA.

But cities aren’t really the area where 5G home internet provides an advantage. Sure, the protocols more efficient than WiFi - the tower allocated bandwidth and time blocks to coordinate transmissions instead of having everyone’s router blast away on the same hardcoded WiFi channels. But it still achieves those high speeds and great connectivity through having a bunch of small cell towers - hardly much different than wired internet.

The real advantage is suburbs, small towns, and rural areas. You solve the last mile problem for cables, eliminate outages, and you do it all with a couple large cell towers.

Yeah, people (or more likely businesses) with high throughput needs will still use wired connections. But it will be because the electricity costs of transmitting from a 5G tower are high enough that there’s always a base cost per GB, not because there isn’t enough bandwidth to go around.

> eliminate outages

Why do you expect 5G to eliminate outages when we can't eliminate them with existing LTE towers, with home wifi, or even with direct ethernet connections to modems?

As of now, my least reliable network is always my cellular network.

> you can have about 100k devices per square kilometer before speeds drop below ideal (20 gigabit)

When, though? We're far, far from this at the moment. What is going to fix it and when?

I think you’re imagining some centralised MMDS type architecture but with cellular architecture this is a reality in many places. Even in modern settings if you have WiFi you need never even be aware that there is a wire there feeding it. Indeed to many of my younger contemporaries it can take a moment to explain the difference ...

I understand there's a horizon for cells (vhf and up isn't going to be reflecting off the ionosphere) but a whole lot of people can exist within a single cell. My argument is about the informational capacity per cell.

Well cells are getting smaller all the time and currently there is enough capacity available in an optimally configured network to provide all the services one could possibly need. Of course you’ve got physical cable tying it all together but the “experience” is wireless.

Each cable node shares ~1GHz of bandwidth among all the separate runs to individual houses. Relative to that, 5G has similar local bandwidth, given the low propagation of mmwave.

It's not going to completely replace hardwired internet in suburbs/cities, but it might just kill last-mile fiber rollout short of new construction.

Not trying to prove you wrong, just asking. Can a service like starlink not deal, in some hypothetical configuration, replace cabled internet?

Can it exist and provide a usable service to some? Yes, probably.

Can it replace cabled internet, period? Can it replace terrestrial cellular? Unlikely:

Mobile internet is constrained by the capacity and coverage of the cell. "Capacity" being the practical sort of what the maximum achievable information rate is, given the available receivers and transmitters. Satellite links are a bit complicated, but essentially the Shannon-Hartley theorem holds and you either have to increase bandwidth or increase signal-to-noise ratio to provide better data rates within a cell. Bigger cell, more users, less capacity per user, worse data rate per user.

I don't know what the cell sizes Starlink will use are, but I've read numbers around 200 km^2. My hometown of 10000 homes is 21 km^2. If they were to provide 100 Mb/s just to my hometown, they would need a 1 Tb/s capacity in a much smaller cell. Of course, not everybody will use 100 Mb/s all the time, but it illustrates the problem.

This is why SpaceX is mainly focusing on low-density areas in places where the infrastructure is less-developed/monopolized. The capacity / coverage problems works much better to their advantage.

Reply on news.ycombinator.com