Why wireless networks are slow | Hacker News Reader

Why wireless networks are slow

mattryall.net

69 pointsmryall20 comments

This may be an incredibly naive question, but is there any reason wireless communication isn't directional, to mitigate some of these interference and square of distance problems? You'd obviously need some lower-bandwidth isotropic mechanism to locate other devices, and then a number of heuristics to track and predict motion.

Perhaps this is already how these things work?

While I'm not an expert in the field, the reason it's not used for standard WiFi networks is probably both that connections would drop out really frequently and that it doesn't scale well (or at all) to large numbers of clients. Directional (line of sight) wireless communication is used in situations where you don't want to or aren't able to lay cables, and have a one-to-one relationship (or one-to-many, as with satellite tv), though.

People have managed to achieve much greater ranges (like more than a kilometre) with 802.11 technology using "cantennas" -- directional antennas like what you're talking about.

While it's theoretically possible, I'm not sure whether this is practical for general use on a LAN. You'd probably have to have a wireless antenna poking out of your laptop, which you'd point at the access point. Seems a little bit inconvenient.

Pointing a big antenna is not how it would work.

You don't need to point the antenna at anything; you can use multiple antennas that steer the beam ("beamforming") in a desired direction. Link:

http://en.wikipedia.org/wiki/Smart_antenna

Basically you can have a "phased array" of N antennas (2 counts as an array, but you can think of a line of say N=8 antennas) that delay their broadcasts relative to each other in such a way that the N signals interfere constructively along some directions and destructively along other directions. This is because, from far away,

  cos(angle_from_line_of_antenna_array) = const*propagation_delay_between_antennas

Ah, clever! Thanks for the link.

AFAIK beamforming is already used in 802.11n.

the military has perfected phase array multi-target tracking technologies quite a long time ago :) There were similar movements in the consumer WiFi arena - that can be equated to hub-to-switch movement that happened in the wired Ethernet years ago - yet somehow it didn't pick up:

http://boingboing.net/2003/02/17/vivato-releases-pric.html

note especially nice thing about beam forming "which allows it to emit at very high, focused power, extending range without running afoul of FCC regs."

We are just now tapping on the spatial dimension of wireless channels; up to now we have just used frequency, time, and polarization for orthogonality. Spatial processing allows theoretically unlimited (bounded by computation, solved by Moore) channel capacity. At the same time, we are beginning to use the available channel resources more efficiently with cognitive radios, which allows much better spectrum utilization. Today, with statically allocated spectrum use, I'd guess that less than 10% of allocated spectrum and time is in active use at any time.

My first thought before even reading the article was: http://www.bufferbloat.net/

Good to see someone else commented about it.

Nice article. The collision avoidance algorithm explanation is very simple, but overall the post is interesting.

Why can't we create more open spectrum?

Just because something isn't open, and is less than perfect, doesn't necessarily mean the solution is to replace it with something open.

No, open spectrum in this case means that's its available for use by consumers (wifi, bluetooth) rather then just companies that get a license from the government (cell, radio)

The feds took over radio spectrum in the early days of spark gap transmitters for the "public good." It's just the way it is unless you want to fight the FCC at this point.

One of the major reasons all the consumer wireless devices suck is that only small slices of junk spectrum was given away to these unlicensed devices (ISM band). Additionally, the chaotic nature of open spectrum requires different contention protocols effect the efficiency and jitter characteristics of wireless networks. When you can't just call the feds to kick people off your band you have to "play nice" is possibly wasteful ways.

The original wifi protocol basically took carrier sense multiple access with collision detection (CSMA/CD) and turned it into collision avoidance (CSMA/CA). Instead of backing of after a collision (which we can't detect in wireless) we back off before every transmission - cutting throughput in half.

because that would require your local government to give up the huge buckets of cash they get for licencing it.

£1bn expectation: http://www.mobilebroadband.net/news/2012-spectrum-auction-pr...

£22.5bn for the UK 3G spectrum: http://webcache.googleusercontent.com/search?q=cache%3AMCpcS...

Would it be illegal to sell firmware that uses more spectrum and boosts wifi speeds?

I'm just wondering how hard it would be to jailbreak WiFi.

It would probably not be illegal to sell the firmware - but it would definitely be illegal to use it.

It's unfortunate there's no mention of using MIMO spatial correlation techniques to increase Performance.

check the end of the article

got to 3gpp.org and read up on how LTE works, many of these problems have been addressed.

Reply on news.ycombinator.com