Product request: The Wifi Antenna switch

[NewBWiFiGuy]

I am not even sure if this is possible. But here is a product I could really use to make my WiFi coverage shine:

A device that either is a complete AP or allows a standard AP to plug into that acts as an antenna switch automatically selecting which antenna in an array to transmit/receive on.

The device would have, say 4, 8, 12.. or heck even 20 antenna ports (standard N connector) with an inline amplifier that a WISP operator could attach highly directional antennas to. Each antenna would then be aimed to cover a diverse or slightly overlapping area enabling a single tower/mountain top to reach a very extended range and coverage area.

The vendor that creates this device also designs an OPEN communication protocall for these Antenna switches to talk to their access points. Over this open protocall this "Antenna Switch" reports which antenna a users signal came in on. The wireless AP records that the MAC address is coming over antenna port #x and upon the next packet it wishes to send to that wireless client it attempts to send it over only that (or adjasently aimed) antenna unit #s.

From then on, all the other ports of course watch for the user as they could be roaming. The next packet in queue to user x who may be in the completely other direction of the last user gets his antenna "switched on" (all this taking place in nano seconds) and the packet is sent to him. And so on and so forth. Picture this, a WISP with a large tower site that happens to have line of site for 100 miles around him because he's on a mountain top (ok I admit, this is me!). This WiFi operator could build a 24 antenna parabolic high gain sector array using two access points running on channels 1 and 6, then two 12 port amplified antenna switches. Just like the latest TV satellites (dishnetwork) the system is now using a spot beam system. This WISP operator would have full overlapping 360deg spot beam coverage that would be able to reach extremely far distances just as if the WISP operator has local antennae located closer to the user. This of course would apply to small installs as well as very large ones such as the example listed above. As long as the antenna is highly directional the system could reach out to far greater distances.

notes;
*The switch should be waterproof and antenna mast mountable and would have the shortest possible cable run to each antenna.

*An AP with 2 completely diverse radio systems (meaning it could operate on channels 1 and 6 or 6 and 12, etc) could feed 2 12way switches for example.

*An AP that is setup to work with one of these switches would automatically send power up the coax to power the switch.

*Since only one antenna would be transmitting at a time there is no risk of overlap, when antenna x is transmitting the switch would shut off the other 11 antenna TX signal. The clients would be kept in sync via a timing signal.

*The protocall should be incredibly simple and logical to implement. Each antenna port comes with a pre assigned ID (or industry wide unique MAC style ID) and with every receive packet it sends a small packet on another frequency via the cable telling the AP what antenna port this user is coming over. Since it's a single directional antenna the user should best receive his next packet over that same antenna. As the user roams the AP would either transmit over the mutlple antennas the user is reaching or migrate their TX/RX transsmissions to antenna X. The protocall standard could be monitored by http://www.wi-fi.org/OpenSection/index.asp

Now before any of you jump on me saying Vivato (http://www.vivato.net) has already invented the "WiFi Switch", you are kind of correct but they use a completely different system and it is proprietary. Plus their pricing is a bit high for me. ($9000 for their indoor unit) I want an under $1000 switch (say for the 8 port) that I can use any standards conforming AP and any antenna. Or a complete product with x number of anteanna outputs that does the job internally. For instance, Linksys/Dlink/Lucent/etc all offer APs with diverse antenna outputs - limit 2. From what I understand, at least the better designed units, figure out which antanna it's got a better connection to the user on and uses that antenna for the packet. Now invision an industrial class AP that sits on your tower doing the same thing but has 12 antenna ports instead.

I'd think if someone develops my idea it shouldn't be difficult to get the top x AP vendors to implement it. The communication standard should be free to all manufactures to build into their APs, but the non-AP switch product itself would be owned by the company. An international standards body would oversee the communication protocall between the switch and the APs. Something to keep it open and expandable so the WISP consumer could have a bit of choice.

Another method would be to have a kind of software radio device that would do all the TX/RX/Switching and a radio less AP module that serves as the brains telling the switch what to transmit. With some flexible design it could be upgradable to the newest standard by replacing the radioless AP or upgrading firmwear.

Idea's? Comments? Am I crazy, is this device impossible to design? It seems logical and incredibly beneficial to the WiFi market. It would allow for small WISP operators to cover extremely larger areas of coverage without buying AP after AP and trying to make sure they didn't interfear with each other.

Thank you.

[ccrum]

Your idea is good but impossible to implement in the way you describe it. A coomon misconception is that if you have a tower top amp, you don't have to worry about the cable run. This is false. Cable is still lossy and potentially noisy. There is still a limit to how much cable you can run and still re-amplify the signal before it degrades too much. Also, a 12 way splitter would destroy the signal. For every split you are taking a 3 db hit or cutting the power in half. You would have vrtually no power at each antenna. Now maybe you are thinking that this would be ok if you could amplify the signal enough before entering the splitter/combiner. It might on the fwd link, but what about the reverse link? Remember that there is nothing you can do about the reverse link received signal level at your location. If you have a crappy signal, putting a bi-directional amp inline is just going to amplify the interfernce and noise coming in as well. Smart antenna systems have failed in almost every application they have been tried in because they simply just don't work. It is very difficult to break the laws of physics. Cellular operators spent a lot of time and money on this kind of thing before finally realizing this. Your best bet is to find a more inexpensive AP. Try smartbridges.com. Their professional quality outdoor APs are nice, compact, waterproof, and under $300 a piece. Beside that, you'll get better throughput using multiple APs. If you design the sectors right, you can minimize the interference from adjacent APs as well.

[NewBWiFiGuy]

Thank you for your reply:
Here are my thoughts.
The cable run issue is a non concern. In my post I said "that has an amplified connection so cable length back to my AP is not an issue" What I should have said is that due to receive and transmit gain of the amp, cable length to the AP shouldn't be a problem if you use the quality stuff and keep your distances low. For instance, 100ft of quality cable with an amp is not a problem.

My thought for the antenna switch is that it would be internally switching to each antenna without standard splitters. This can be done just as the many APs on the market employ diversity antenna. Basically think of it as a direct pipe or connection to each antenna at nanosecond switching speeds. The AP when TX/RX is only connected to 1 antenna (or maybe 2 or 3 if a roaming system was built in, but hey - I'm talking first generation here). So, your point of there being loss like a standard RF splitter would not be a problem because the antenna switch is doing exactly what it's name describes. It is switching. Just like a Hub vs a Switch on a network. A hub tends to broadcast everything to everyone. A good switch is just like having a crossover cable between machine A and B for the individual packet. Nothing else sees it or has to deal with it.

Thanks for your tips on the network design with multiple APs. That's pretty much what I've got laid out right now. If someone could design this product however, I could still use multiple APs but cover a much larger distance..

At least that is the theory. ;-)

[NewBWiFiGuy]

Update for this thread: I received a couple messages from someone who is apparently a pretty good techie that MikroTik with their RouterOS software (a small linux appliance linux build build to route and do lots of tricks) that it is likely MikroTik could implement this in their software. For instance, a small embedded linux box built to live before your multi access points that would be confiurable and trainable via an intelligent algorithm to only send packets to directionally located users. In other words, say you have 4 APs on channel 1, they'd be divided up by the software to only transmit one AP at a time.

This way, very high gain long range spot beam coverage antenna systems could be designed. WISP operators that have really high points (like mountains and upper eleviation) that need to be able to cover huge amounts of area using diverse radio systems could be helped by such a system. It would be allot slower than a hardware antenna swtich/system but it would do the job for a very low cost.

Also it was suggested that a small linux MikroTik RouterOS box could be built into a weatherproof enclosure that would sit on your tower with say 5 pci 802.11 radios in it that could create a big multichannel coverage system that would diversely cover a larger area of people by slightly slowing down the packet delivery and acceptance/RX rate.

Who knows folks.. it would work, and you'd be able cover a longer diameter reception range per antenna.. but it would be a little slower.

[wispy]

Have looked with interest at MikroTik RouterOS implementation of a low cost Linux Box (with 6 or more PCI slots) as a possible multiple sector antenna to simulate (albeit a much slower simulation) the PhasedArray antenna of Vivato.

See: http://www.mikrotik.lv/Documentation/manual_2.5/Interface/Prism.html

The interesting thing in this doc is their statement: "The MikroTik RouterOS supports as many PrismII chipset based cards on as many free resources as are on your system, i.e., IRQs and adapter slots. One license is valid for all cards on your system. "

Thus, using their Prism 2.5 cards, one could have 6 or more slots for cards to give full controllable/adjustable power to each sector antenna up to 200 Mw.

RouterOS v2.6 is available on a 64MB IDE Flash Module for only $170. 200 Mw Prism 2.5 cards with PCI adapter are $175. You can easily calculate the cost (don't forget antenna cost).

Anyone else out there using RouterOS?

[dastormy1]

There is a product that does what you're looking for ... it's basically a switch ... and you run antennas to each port ... I can't go into too much more detail since the company is still in stealth mode, but the product allows the antennas to capture the 802.11 packets ... and then everything is handled at the switch.

This allows the antennas to all operate on the same channel since all the magic happens at the switch. The switch can support up to 60 antennas ... and the antennas are 1/4th the cost of APs.

Basically ... this removes the headache of having to manage a bunch of APs. It's a great product.

If you want more information ... let me know and we can take this offline.

[NewBWiFiGuy]

How exciting.. See I knew my idea wasn't new.. I'll send you my contact info offline.

Of course as I said before, the switch has to be low cost - otherwise the savings on access points benefit is canceled out. It also has to support long range hotzones, short range in building zones are not of concern to me. Those tend to be pretty easy anyway using standard APs.

Note my vision of having a moutain top major AP hub. Picture a mesh of long range point to multipoint antennas covering very wide areas. Just like spot beam satellites work. A single site intelligently transmitting powerful directed beams covering x number of sq miles with each beam. If my mountain top can see xxx sq miles I would design the switched antennas to each beam their coverage over a percentage of that target area. Since 802.11b only has 3 non-overlapping channels, without a switch this idea is very difficult.

[NewBWiFiGuy]

You aren't the first person to mention a software solution such as MikroTik. Since MikroTik is still not a turn key provider like the big boys, I'd still want to use someone like Pronto/Airpath/etc at the AAA controller headend for the billing, control, and customer support pieces. BUT, if MikroTik could provide edge of network solutions that prove themselves then I am game.

A software switch would be slower, but technically feasable, cheap, and very configurable I'd think. With enough work it could do some impressive things.

Since a PC could only hold 4 or 5 cards each, perhaps the software could send out a timing/sync signal to other slave PCs over a dedicated interface to make sure signals were not overlapping. In other words, picture 3 PCs with 5 radios each all operating on channel 6 with diverse directional antennas. You'd be able to aim out 15 high gain antennas in different directions to cover a very wide area. Sure this is a jury rig method, but, it uses all off the shelf parts and a single 5 radio box could be built for under $1500 in parts. 15 antennas, cable, connectors might run you $2000-$2500.

I wonder how many ms latency your going to add to each packet with a design like this. 5-10ms? More? Less?