Intel Wi-Fi Provides 6 Mbps Over 100 km

MIT Technology Review describes a new Wi-Fi router from Intel capable of sending a Wi-Fi signal tens of miles with 6-Mbps performance. This is perfect for rural areas without Internet service, and for less developed countries interested in building out their Internet infrastructure but no means to lay expensive cable or fiber optics. The routers cost about $500 each, and you need two of them for a point-to-point connection. Quoting: "Intel's RCP platform rewrites the communication rules of Wi-Fi radios. Galinvosky explains that the software creates specific time slots in which each of the two radios listens and talks, so there's no extra data being sent confirming transmissions. 'We're not taking up all the bandwidth waiting for acknowledgments,' he says. Since there is an inherent trade-off between the amount of available bandwidth and the distance that a signal can travel, the more bandwidth is available, the farther a signal can travel."

Comments

still too expensive

[Lumpy]

When a pair of linksys routers, 2 old and free Dish network dishes and $30.00 worth of parts can to the exact same thing.

Even if they were available when I helped start a community wifi, we would not use them. they are too expensive. We are getting WRT54GL routers for $50.00 each, and tere is a never ending supply of free dish network dish assemblies with mounts.

Re:still too expensive

[morgan_greywolf]

Yup. You can often get Dish network dishes from Freecycle, and worse comes to worse, I've seen them at yard sales for as little as $5 or so. Throw DD-WRT on a pair of Linksys routers, get the dishes and then follow these or similar instructions and there you go. The whole deal will cost you under $200.

Re:still too expensive

[Jon_S]

I know this is /. and you aren't supposed to RTFA, but I did anyway. It's not just the hardware. They redid the commnication protocols from scratch also that greatly increases the speed over these distances. Your Linksys routers will still be doing regular wi-fi.

Re:still too expensive

[billcopc]

It's far more complex than what you suggest. I'll quote from the FCC:

(b)The maximum peak output power of the intentional radiator shall not exceed the following:
(1) For frequency hopping systems operating in the 2400-2483.5 MHz or 5725-5850 MHz band and for all direct sequence systems: 1 watt.

(2) For frequency hopping systems operating in the 902-928 MHz band: 1 watt for systems employing at least 50 hopping channels; and, 0.25 watts for systems employing less than 50 hopping channels, but at least 25 hopping channels, as permitted under paragraph (a)(1)(i) of this section.

(3) Except as shown in paragraphs (b)(3) (i), (ii) and (iii) of this section, if transmitting antennas of directional gain greater than 6 dBi are used the peak output power from the intentional radiator shall be reduced below the stated values in paragraphs (b)(1) or (b)(2) of this section, as appropriate, by the amount in dB that the directional gain of the antenna exceeds 6 dBi.

(i) Systems operating in the 2400-2483.5 MHz band that are used exclusively for fixed, point-to-point operations may employ transmitting antennas with directional gain greater than 6 dBi provided the maximum peak output power of the intentional radiator is reduced by 1 dB for every 3 dB that the directional gain of the antenna exceeds 6 dBi."

Now, is the FCC going to troll around your neighborhood with a scanner ? Probably not, unless you screw up someone else's wireless equipment. Done properly, a high-power point-to-point system shouldn't affect anyone else, so you can probably go nuts. I can't say, I don't even live in the US, but my guess is the intent of the FCC regulation is to prevent, or at least document, people from stomping all over the spectrum with uber amplifiers. If it weren't for such rules, inevitably someone would create a 20-watt cordless phone that fries small birds but gets killer range - and also clobbers everyone else's phones.

Perfect for regional australia

[awdau]

There are so many areas within range of regional cities that only have dialup.

Re:Perfect for regional australia

[CastrTroy]

Yes, but that's 6 mbps split between everybody who wants to use it. Let's say you have a small town with 400 computers. And lets say that 1/4 of them want to go online at the same time. So, we have 6 mbps / 100 users, and you have 60 kbps per user. Which ends up being not that much faster than your average dial-up service. Using wireless is like hooking everybody up to a single hub. The bandwidth gets shared between all the users. Works great when you have 4 people sharing a 10 mbit LAN connection to a 1 mbit internet connection, but not so well when you want hundreds of people on the same network.

Re:Perfect for regional australia

[rmadmin]

Actually, bandwidth scales much better than this. I used to run 600~ Cable modems at 256Kbps on 4 T1s. That happens to work out to 6Mbit. Towards the end we were peaking often, but overall it worked ok. 400 Users at 200Kbps is reasonable in my experience. And 200Kbps is far better than dialup if it's your only choice. =)

A lot of issues with this

[transporter_ii]

Didn't read TFA yet, but I know this will work fine with two units, you just set one to provide sync. But if you have four units in an area, they can interfere with each other. What you can do then is add a gps unit to the AP side, sync to that, and all four units Tx/Rx at the same time. So MIT really just created a Wi-Fi Canopy system...or what WiMax will be if it is ever released.

The biggest issue is that 2.4, with only 3 non-overlaping channels, is it almost unusable for long distance shots. I'm working in a WISP that has some 2.4 and it will make you pull your hair out. At one tower, in somewhat of a rural area, we could see 121 different SSIDs from an omni antenna a couple of hundred feet off the ground.

At 500.00 a unit, I doubt this will see high deployment, but if all of these things don't play nice with each other, it will be yet more interference.

And last, 2.4 could already do ten miles easy already, and much cheaper. You could build a Mikrotik AP for 600.00ish and have 20 clients at 10 miles for 200ish a client unit, if they are all line of sight. But note that you have stretched 2.4 well beyound what it was designed for, and in no time you will understand exactly why WISPs startup and fold like crazy...and the only people who made ANY money are the ones who sold you the equipment.

Transporter_ii

Re:A lot of issues with this

[transporter_ii]

The problem with Canopy, as it is designed, is that the tower sites cost a fortune because the APs only have 60 degrees of coverage and, as designed, it would take 6 APs (900.00 to 1,500.00 each) to have 360 degree coverage. But it is possible to connectorize the APs and use far less APs by adding antennas with more coverage. If you can stomach that, you can get client units for less than 300.00. And add an aftermarket sync unit (200.00 - 300.00), and you can have multiple APs and clients not interfere with each other (just hope the other guy in town with Canopy syncs his equipment).

Let me tell you, two to three times, I have been involved in a 2.4 build-out. Each time it went like this. You spend a lot of time and money going around and swapping out that "expensive" Canopy equipment for the much cheaper 2.4 equipment. Everything works fine for about four days to a week. You run back and swap a few people back to that "expensive" Canopy equipment for various reasons...but within six months, when the crap hits the fan for some reason, and you have to have help scrambling to find enough Canopy equipment to put everyone back on...because its the only thing that "just works." It may not be perfect, but it does work.

After it saves your ass a few times, that Canopy equipment doesn't seem so expensive.

Transporter_ii

Perfect.....

[rolfwind]

if you don't require privacy. Hopefully, they put in an extremely good encryption scheme with this and not one merely "good 'nough'. Still a good leap forward in many areas, our country is way behind as it is, and it has next to nothing to do with population density for the east and west coasts many areas of which has poor, overpriced service as well.

I often wondered what is stopping a mesh network from spreading. It would be basically the type which the OLPC has, except essentially a router with an antenna could be put on top of your house and connect with others of its type, from spreading. Of course, there would have to be a central hub connected into a fat pipe every so often so the signal doesn't hop around like mad.

Re:Perfect.....

[Agripa]

I often wondered what is stopping a mesh network from spreading. It would be basically the type which the OLPC has, except essentially a router with an antenna could be put on top of your house and connect with others of its type, from spreading.

I have done a little work on this problem over the years and I suspect there is just a lack of all the necessary pieces for a good high performance mesh network solution. Here are some ideas off the top of my head while ignoring economic and political reasons:

1. Current radio hardware and band allocations only support half-duplex communications. WiMax uses transmit and receive synchronization to lower the dead time and prevent collisions which helps but how do you synchronize an arbitrary number of half-duplex stations in a variable geometry environment without a significant loss of throughput?

2. How well does IP handle a constantly changing network topology with hidden nodes? I suspect overlaying IP onto a protocol specifically designed to handle routing in an adverse wireless environment would help. Every node should maintain an extensive situational awareness of its local routing environment to provide for instant failover and redundant routing.

3. QoS would require some type of sharing scheme that does not rely on the good intentions of every node. BitTorrent accomplishes this using Tit-for-Tat. IP accomplishes this by using flow control and assuming a largely benign network. While computationally expensive, I suspect some type of cryptography based token scheme would allow both trust metrics and something like a packet routing barter system. Notice that this automatically allows the client to assign priorities to different types of traffic while intermediate nodes can accept the client's word given enough trust.

The catch

[A beautiful mind]

The connection can only be established between two nuclear power stations.

Bad article summary

[j.a.mcguire]

This is poorly summed up, the point of this is not the range or the speed, its the fact that it only uses 6Watts firing data at that range and speed and could use stand alone, solar powered units to maintain data links.

Single point failure.

[ruin20]

I go back to the first poster alternative about cheaper alternatives, I've seen some extremely interesting work with mesh networks, and they provide a level of redundancy not present in this system. And that's important if your going to talk Canopy or WiMax or something because now your talking about infrastructure. If you have one tower covering this kind of range imagine the amount of customers a failure effects. We can create mesh networks with existing technology and for a lot less money.

They Never Heard of ZModem?

[Toad-san]

"Galinvosky explains that the software creates specific time slots in which each of the two radios listens and talks, so there's no extra data being sent confirming transmissions. 'We're not taking up all the bandwidth waiting for acknowledgments,' he says."

Doh .. Huge difference in the later early modem data transfer protocols was (1) variable packet size (if noise went up, packet size would drop down) and (most important): No ACK/NAK! Sender just sent as fast as its little chips could push the data out. Receiver would just receive and stuff the data away. It was only when the receiver did NOT get a good packet that it would do a NAK (and send the number of the bad / required packet). The sender would stop what it was doing, drop back to the bad packet number, and retransmit from there. (With more memory and speed, it would've been better to buffer packets so sender only had to send the single bad packet, and then could resume where it was further down the data stream. But I digress.)

So signal conditions are so lousy with wireless data transmit protocols that they're still doing ACK/NAK for every single steenking packet? That's pretty dumb, eh?

Toad-san