Say Bye-Bye to Interference, Really ?
I saw a write up of Extricom's (www.extricom.com) product offering being implemented at the University of Georgia Athletic Department on WiFiplanet.com and wanted to see what the real scoop on Extricom was. (http://www.wi-fiplanet.com/news/article.php/3667156)
It appears that they have come up with a way to bypass co-channel interference. I was told that the University of Georgia IT department, a Cisco shop, was shown the Extricom product and were blown away. This company has seemingly rewritten wifi deployment by removing the necessity for site surveys (mostly) and cell planning.
How do they address non-extricom equipment, won't there still be interference? What happens in an apartment building type scenario when everyone has their own personal wifi and the airwaves are flooded? How does a sending station know that the carrier is available to send when another user may be so far away that the signal is not heard?
Please let me know if this is really something that is viable. The company deploying for UGA says that it is a simple to manage and deploy solution and they have succesfully deployed the system already with more indoor and outdoor deployments to come, including Sanford Stadium. Anyone care to comment?
I talked to a salesperson at some length today. While he was light on technical details, he had a good story, and said they have a lot of patents issued or pending.
I downloaded some whitepapers and network media reports to see how real it might be.
If I come up with any hard data, I'll post it.
Sadly, the cheapest test system is $10K, for a controller and 8 "APs".
The article states that they are using the same channel for every AP. Hmm.. can someone say Meru?
Ever since hearing of Meru I have theorized that Meru (and now others) compensate for AP's on the same channel by timing transmissions from the AP's. Meru's use of CTS frames to clear the channel has been documented although they called this an "error".
Beyond a well placed and well timed CTS frame, I also theorize that these manufacturers could negate normal arbitration between the AP's that are on the same channel.
Imagine this; A CTS frame is sent with a duration of 500 microseconds (made up number that would equal all SIFS, ACKs and DATA frames to come from collocated AP's). AP's 1, 2 and 3 are co-located all on the same channel. The controller could then send data to AP's 1, then 2 and then 3 and tell them not to arbitrate. The first AP transmits, then the second AP waits a SIFS + ACK + SIFS before transmitting its own frame as opposed to SIFS + ACK + DIFS + Back off.
This is completely made up by me for whatever that is worth. However, it could theoretically improve performance.
Thanks Charles and GT for your input. I will be visited by the Extricom rep next week and will be presenting him with some pretty technical questions. I don't know if the sales rep can answer these, but at least he can refer them back to a tech who can. I don't imagine they are going to tell me exactly how it works, but at least we can theoretically understand the process.
If anyone has any questions they would like me to pose to the rep, please send them on. I will post the response once I get them.
Well, I don't quite undestand this. DIFS and backoff "naturally" happens on a wireless network. Let us say client1 is done with the data transfer after RTS, SIFS, CTS and whole nine yards. Client2 will obviously going to wait for DIFS before sending it's own RTS isn't it. Their controller can control APs but I don't think they can control client's behavior.
I am very interested in knowing how they are doing it from their Marketing guy.
The rep had to reschedule for next week. If you have any questions to ask, let me know on the forum and I will find out.
The information I was given about the design history is that they were specifically intending to prevent a co-channel interference meltdown that would occur with lots of separate networks or dense AP deployment in a limited area. One of their selling points is that they can go to existing deployments with coverage holes and put in a controller with 2-8 radios, which will seamlessly co-exist, and improve the coverage.
Their salesperson either stated or implied (or I inferred) that they make a transmit choice per radio on a frame by frame basis. Since they have a choice of which radio per frame per client, they could use a radio that's ok for the client while being more distant from the current instantaneous co-channel interferring source. (Caution - wild speculation warning!).
If I had a spare $10K for their minimum controller configuration, maybe testing would make this more understandable.
Okay, so we still have to contend with interference. The spectrum is what it is and the Extricom switch does not get rid of interference when it is in an environment containing or bordering an existing wireless infrastructure.
What it does do, however, is, in a native Extricom environment, essentially eliminate co-channel interference. The biggest adjustment to me is getting my brain wrapped around the idea of replacing a cellular WiFi network with "One Big AP". The Access Points are replaced with "Radio Points" that transmit layer 2 data back to the switch which does all the processing and determines where the packets should be sent, much like a wired switch.
When I asked how it does this, the response I got was that it was "the secret is in the sauce". In other words, I was not going to get an answer. However, there are patents for the Extricom that can be viewed publicly according to the sales rep.
- Switch is easily configured. The interface is very simple, without the need for setting channel and output power, there is not much left.
- Supports Latest security (AES/WPA2)
- Any updates to the firmware on the switch happen in one place. Reboots are very quick.
- Radio Points boot up in about 3 seconds
- Supports up to 16 VLANS
- Multiple SSIDs w/ Multiple Encryption with or without VLANs
- Radio Points can be located up to 600 ft w/ proprietary POE or farther with Fiber
- Radio Point turns off when not in use
- Since there is only one "Big AP", your network only looks like it has one AP when viewed in netstumbler, etc.
The drawbacks to the switch are as follows:
- Unit cannot be used in an environment with other existing wireless. You would have to remove an existing infrastructure and replace it with this one.
- Unit must be on the same subnet (this can be addressed by putting unit on a VLAN with other Extricom devices)
The bottom line:
The unit does what it says it can do. If you have a new environment that needs wireless or if you have an environment where people have installed a haphazard wireless network, this is the way to go.
We are going to continue testing the unit in house as it has been left with us for testing. Anyone that wants to remote in and take a look is welcome, just private message me and I'll set it up. We have the unit for the rest of the week.
The Cost: MSRP for the switch is about $6000 and the RPs are about $300. Note that I quoted MSRP, actual prices will vary.