Hacker News Reader

Those antennas are just for show.

There are plenty of short-range wireless technologies currently under development that use the SHF/EHF bands to send data at very high speeds. There is plenty of bandwidth up at 60GHz. Furthermore, the processing power of affordable chips is very high, so they can control an array of antennas that send the signal exactly where the other end of the connection is (WiFi is starting to do this, and cellular networks have done this for a while). For connecting your hard drive and monitor to a nearby laptop, I think this will work fine.

The big "problem" with these bands is that they don't propagate through walls, or even your hand or a sheet of paper. But with beamforming they can go around your hand, so it's not a problem for things like connecting to monitors, in theory.

Wall penetration is a big problem for WiFi. WiFi's very limited bandwidth is shared among all stations that can hear each other. When walls block the signal, it's like you're creating a brand new Universe free of other stations and interference, and so you get free bandwidth. This is why corporate AP deployments turn _down_ the power of their access points, so they only cover a small area, thus allowing the system to have more total bandwidth.

Ideally the clients would also turn down their transmit power, but they don't because that's bad for benchmarks and reviews, so they kind of ruin this by bleeding their signals intended for the nearby AP into the range where other clients that can't hear that AP are. Fortunately things like beamforming and MU-MIMO are fixing this, by allowing receivers to move a null in their receive antenna system to where the unwanted device is, thus providing isolation from nearby devices and increasing the total bandwidth of the system. We've been promised this for years but it's still in what I think are the early stages.

(As for the use case of "I want one WiFi router to cover my whole city!" we're not there yet. You need an AP in every room, with a wired backhaul. Meshing might solve this... but it is also in the very early stages.)

The antennas are not just for show. Having multiple antennas allows you to increase bandwidth, reliability and range.

https://en.wikipedia.org/wiki/MIMO

https://en.wikipedia.org/wiki/Antenna_diversity

https://en.wikipedia.org/wiki/Beamforming

The post I replied to linked to this: http://www.amazon.com/dp/B016EWKQAQ

You do not need your antennas to look like that, that's pure marketing.

Here's an access point with 12 antennas: https://on.google.com/hub/

Oh you're right you can pack them more discreet, but that should have a worse performance than the one I showed. That's because the coherence length of wireless is fundamentally related to the wavelength -- so to get independent signals you need a certain physical separation, no matter how clever you are with their placing.

Antenna length and placement certainly is a huge factor in wireless performance. But those specific antennas on the linked router are thick and huge mostly because of the plastic decoration around them. As the GP said, it's marketing.

The wavelength is 6cm for 5GHz and 12cm for 2.4GHz, so a 1/2 wave dipole will be half that. (Nobody uses dipoles for WiFi, though. Undesirable radiation pattern, etc.)

From what I understand, beamforming won't allow high frequencies like 60GHz to 'bend it like beckham'. If your hand can block 60GHz, then the radio waves probably aren't bouncing around the room enough for it to make a difference. This is because, unlike the name, beamforming is simply phase modulation used to alter the constructive and deconstruction interference patterns of the broadcast radio waves. So for 2.4GHz, where you already have radio waves coming in multiple directions, the constructive interference patterns can give you a higher signal strength.

> "Eventually, most of these connections will likely become wireless." ... I have my doubts.

Eventually. Not tomorrow. Not next year. But it will come. If you include the full quote, it becomes in my opinion clearer what he thinks:

> Eventually, most of these connections will likely become wireless, but given the need for power and the expected challenges around delivering wireless power to many devices, it’s clear variations on USB Type-C, particularly Thunderbolt 3.0 and later iterations, will be around for some time to come.

While there are physical limitations to what wireless can do, over the last decades we've had groundbreaking research which repeatedly has increased the BW available to wireless systems by orders of magnitudes.

Why do you think we've hit the peak now and it can't happen again?

> The proliferation of USB Type-C clearly marks the dawn of a great new era of connectivity for our devices, but it may require a bit of homework on your part to fully enjoy it.

Fully agreed on this. In my head USB Type-C was ... just USB Type-C until I read this article.

>Eventually. Not tomorrow. Not next year. But it will come.

Physics and physical limits don't really change with time.

But naive methods give way to more clever ones.

Yes, but clever methods can't bypass physical limits either. Sometimes that's just it.

Bypassing limits is exactly what clever solutions do. We can't break physics, but we can use different methods to solve the same problem. Like wires to get into a room, and a lifi or laser network inside of it, or wireless between the rooms with furniture sized antenna. Or, more likely, something far more clever.

> While there are physical limitations to what wireless can do, over the last decades we've had groundbreaking research which repeatedly has increased the BW available to wireless systems by orders of magnitudes.

And they have been outpaced by the bandwidth improvements on wired networking.

Why settle for some increase in speed (maybe, assuming you don't have signal quality problems – which only grow worse as adoption increases), when you can have a much larger increase in speed (guaranteed)?

Sure, wired doesn't always make sense: Anything that can be sensibly battery powered is a candidate for wireless connectivity. But if you're stringing wires anyway, why not use much more reliable wired networking while you're at it? We're putting power delivery into Ethernet and USB, and Ethernet into HDMI, so we're increasingly not even needing additional wires.

I am not an adept of wireless because of performance problems but wire consolidation makes a lot of sense. For instance it would make sense for all laptops to have the ethernet connection as well as additional connections going through the power supply. So that a single wire to plug when you go wired and all wireless when on the move.

Docking stations sort of do that but you loose the "lap top" ability. I think most people use their laptop in bed or on a couch when not working. And even on a table it's nice to have the wiring underneath the table.

> So that a single wire to plug when you go wired and all wireless when on the move.

That's the whole idea behind USB-C with Thunderbolt 3 and USB 3.1, yes.

Thunderbolt can accomodate many protocoles but can it do them at the same time? I.e. drive a 4k monitor, and an ethernet connection, and power the laptop, and connect an external hard drive and printer through USB? To go the power supply hub way you need simultaneity.

That's not a half-bad idea, you could do that relatively easily with current-gen tech by building a powerline-ethernet adapter into the charger brick. You might need to use a "dock" connector that has pins broken out for the ethernet.

Powerline Ethernet has some issues, its bandwidth is limited and it can be susceptible to interference. But when you can't run a real ethernet cable, it's a very solid option. In particular it's much lower latency than wifi.

I would used a custom connector for the power adapted. Powerline ethernet is slow and messy. Pretty much any laptop uses a custom port anyway so that wouldn't be a problem in itself.

In fact Surface does pretty much that: https://www.microsoft.com/surface/en-us/accessories/surface-... except that it comes with a very short cable which makes it unsuitable for sofas or beds.

Again, powerline ethernet isn't ideal. It's two selling points are (1) not having to run a new cable, and (2) significantly improved latency and packet loss compared to wifi while maintaining reasonable bandwidth. I normally get about 70-80 mbps across my PLE adapters. It is susceptible to picking up some interference at times, which cuts performance in half, but during those times the wifi is unusable for any sort of realtime/latency-sensitive applications. I have local GigE segments in my living room and computer room, but I would need to run a drop to connect them and I rent.

USB-C and Thunderbolt are point-to-point bus connectors, they're not designed to have multiple clients on a network. Even if you designed a switch that could use ethernet-over-Type-C as a physical layer, the maximum length for an active cable is only 50 feet for USB 3.0 speeds and 82 feet for 2.0 speeds. Passive cables are 15ft. That's not really enough for long runs inside a wall, so we're back to asking the question: OK, I plugged my USB Type-C into the wall. Now what?

If you're going to actually run a new cable to your outlets, running Type-C seems like a poor choice. You could stick a Type C ethernet adapter in a wall-socket formfactor, run CAT-6 in the walls, and call it a day. That gets you runs of up to 330ft for GigE. But if your vision is "plug into the wall and instantly get on my network, without pulling ethernet to every outlet" then you need something that runs over a power-line, because that's the only wire that's guaranteed to be run to a power outlet.

Again, not powerline ethernet.

You can create a custom connector for your power plug (power plug to device) with 40-ish pins that make all the individual wires go through one cable (2 pins for power, 8 for ethernet, 10 for USB, 19 for HDMI, and I am sure there are synergies).

Then the power supply has an ethernet socket, USB socket, etc, and all your accessories (printer, TV, etc) are connected to the power supply instead of connected to the device.

The surface dock does effectively that, except that I suspect they only pass a USB3 wire along the power, and have a USB switch inside the dock, and that ethernet goes over USB instead of directly.

Wireless, pff. Give me a wire any day.

At any given technology stage a wired connection will be faster than wireless. Not to mention you don't have to worry about interference, and losing the signal.

If wireless is so great, tell me this, how come wireless routers have wires _inside_ them! :-)

Li-Fi can be extremely high bandwidth.

https://en.wikipedia.org/wiki/Li-Fi

"It is so far measured to be about 100 times faster than some Wi-Fi implementations, reaching speeds of 224 gigabits per second."

I know that in practice, my colleague has tried it, and yes, in practice, it can work.

Only works with line of sight though, not useful or practical when in most line sight situations you could just run a cable anyways, and not have to worry about potential interference.

Seems pretty practical for e.g. a docking station scenario - rather than connecting a cable, just toss your phone on the desk and it connects to your monitor/docking station.

There are multiple problems with LiFi. The most obvious being diffusion and diffraction as well as line of sight (you need a bright light very close to your device). But also LiFi relies on there being essentially zero power fluctuation in the actual lights. This might be possible in a tech demo where the lights were on for 5 minutes, but not for lights that have seen several months of use.

Not to mention that the power costs are immense (only a fraction of the power of LiFi is actually spent on data, the rest is actually powering an array of lights).

> http://www.amazon.com/dp/B016EWKQAQ

Hell with wifi, I just want one of those to hang on my wall as a sci-fi art piece.