Great job GT. Excellent presentation of a commonly overlooked problem.

Tom Carpenter

I understand that Ruckus uses beam forming to direct higher gain to focus better performance to wireless clients. When this happens are you not creating hidden node problems / hidden AP problems because now all wireless clients associated to that AP cannot see the transmission and could decide to transmit. Whats the technology you then use to ameliorate that problem ?.

Tom,

Thank you for the nice comments!

Bells,

You are correct, you can create hidden node (hidden STA as I call it) problem. However, the extra signal that is achieved from the Ruckus AP isn't, in my mind, for extra range although that is nice. One main advantage is higher data rates for each transmission.

Having a directed beam can cause hidden STA problem, but the throughput results are actually better with the beams in part due to the fact that when the AP TX a frame, that frame is in the air a lot less time (due to the much higher data rate). Less time in the air reduces the chance of hidden STA corruption.

Not sure if you watched the video in the blog, but hidden AP problem is a much worse problem today. In the video I talked about how RTS/CTS could, in many cases, help solve that problem. However, what if the AP, when transmitting to a STA could send the signal just to that STA and not in an omni-directional pattern? Hidden AP problem is virtually eliminated.

GT

What you miss is that the real collision domain is the area covered by each AP plus the transmit range of each and every clients in the cell (that spills well outside the AP coverage area). Beamforming or sectorization greatly helps, but what helps more is Multi-Channel beamforming or a Multi-Channel sectorized system. See Xirrus'whitepaper "Benefits of a Multi-channel "sectorization.

Hi GT what you miss is that clients are still having omni antennas...

Wow... I guess I'm missing a lot these days.

Both of you point out the same thing and I'm pretty confident that I'm not missing it. Yes, all STA's have omni antennas and the collision domain does extend outside of the AP's. And... there isn't anything you can do about it either.

One key to all of this is the direction of traffic. In most Wi-Fi networks, 80 % of all traffic comes from the AP to STA. There are very few exceptions to this.

What we can control and what needs to be improved is the data rate from the AP to the STA, and all types of beam forming (static (Xirrus), chip based (Cisco), and Dynamic (Ruckus)) can do that. However, just increasing signal is counter-productive, otherwise we would just jack up the power, install higher gain antennas or just install an AP every 10 sq ft. Bad ju ju.

What the video is pointing out is that in a three (and four) channel design, AP's WILL cause collisions because they can't hear each other but there are common STA's on the same channel, but associated to two different AP's. Solving this problem will significantly improve throughput and reliability in many networks.

I think this is a good discussion and maybe we should move it to the forums where more people would keep up with it.

GT

Ok let's move to the forum. I understand it's tricky for you to have marketing messages match technical reality. You can't just claim that 80% of the traffic is usually from AP to client. At least not in enterprise, cause you're running many different services not just pure data or multicast. You're closer to 60/65 than 80...

so where are the forums ?.

You are esentially saying that the AP must temporarily change its antennae pattern to focus on one client. By doing this then the other AP becomes even more hidden if that other AP is inconviniently not in the same exact place as the client you are focusing your antenna on

Secondly you create a bigger hiddne problem. By having the ap focus its antenna pattern to focus on one client then whilst it is transmitting them all other clients cannot see it is transmitting and all other APs cannot see it.

So you are going to have a big hidden rpoblem. You must therefore mitigate this either by transmitting RTS CTS all the time (uses up bandwitdh) or reserving channle time.

Dont get me wrong i like the focused beam thing but you cant claim it improves hidden problems when it actually makes them worse.

You can find the forums from the homepage by hovering over the Learning Resources drop down, or use the link to go directly to this forum thread:

http://www.cwnp.com/phpBB2/vie...8498#18498

Before we move the technical part of this conversation entirely to the forum, CWNP (and our partners, such as GT) takes many pains to avoid marketing messages that don't line up with 'technical reality.' GT's video isn't supposed to be an all-inclusive treaty on collision domains and the ways to solve those problems. He's bringing awareness to a common problem that is commonly misunderstood, or unknown altogether. Similarly, GT's reference to 80% is not a hard and fast rule and I'm sure he'd readily admit that (see the forum post). The point was that most traffic is downlink... each environment is different, so Tony's network may be 60/65, many university environments are probably higher than 80%, and others will be some other number. The key here is that in most networks, downlink traffic is a notable majority, which Ruckus' solution happens to handle nicely. We certainly wish that we could spew marketing messages that don't line up with technical reality because we'd make more money that way. We are continually telling vendors that we won't defend their solution and marketing claims if they don't jive with the technology. We reject a lot of business because of that. Sorry for my rant, but we work hard to keep unfounded technical crap out, which is why GT has been happy to defend the technical merits of his claims. Thanks for all the input. I know everyone wants to keep CWNP on the straight and narrow and we appreciate that. We want that too and will readily admit it if we've said something that is wrong. Sorry for my defensiveness here, but GT's claims are not, and will never be, marketing messages devoid of technical veracity.

Feel free to continue to post comments and criticisms here, but the technical polemic will be moved to the forum. Thanks again for the vigorous participation. Good times. :)

We need to appreciate that GT has brought up the hidden AP problem which most of US are not Aware of.