Carl Weinschenk spoke with Michael Finneran, the president of dBrn Associates.
Weinschenk: So we finally have a fully fledged 802.11n standard. How did the process go?
Finneran: The process was a little slow. The Wi-Fi Alliance did a superb job, plain and simple. They certified for backwards compatibility just like they promised. We knew it was coming, and it came. We don't look at ratification as a big thing at all.
"The buying community does not seem to recognize that the biggest change, the biggest advantage, is transmission chains."
Weinschenk: What does the movement from "draft n" to full standard mean to users?
Finneran: Just about nothing. The upgrade is not even going to require a firmware upgrade. With the ratification, the big news is nothing major has changed. The big story came a couple of years ago.
Weinschenk: What is the biggest difference between 802.11n and previous standards?
Finneran: The biggest difference is the transmission rates. You can work in either in the 2.4 or 5 GHz bands, in 20 or 40 MHz channels. The 20 MHz channel is roughly six times as fast as "g." It's about 279 million bits per second. In a 40 MHz channel, it's about 600 million bits per second, though those rates are not deliverable today. They are limited in real products. The determining factor is the number of transmission chains that are implemented. .n allows you to multiply the number. That's the big factor brought by MIMO. The standard and draft standard offer as many as four transmission chains.
Weinschenk: What is a "transmission chain?"
Finneran: Essentially, MIMO allows you send multiple transmissions simultaneously on the same channel. If I am using a four-chain system, I divide the transmission by four and send a quarter in each of the signals. Usually, if you send four transmissions in four channels, they would interfere with each other and all is lost. But by spacing them apart, each is received and recognized independently. That is the key. It allows you to have one channel that is treated like two, three or four channels.
Weinschenk: How many of these chains are there?
Finneran: The standard allows four transmission chains, but today the limit is two or three. The buying community does not seem to recognize that the biggest change, the biggest advantage, is transmission chains. They think if you buy "n" you have top of the line. That is not the case. You have to see how it is implemented in what you buy. Otherwise, it is a pig in a poke. You don't know what you are buying.
Weinschenk: Can you start with a maximum capability of three chains and move to four when it becomes available?
Finneran: You would have to change the device to get the advantage. It is not a software upgrade or something simple like that. Whatever product you buy today, it will have a certain number of chains. What you got is what you got. In planning, you have to determine what you are buying and what capacity implementation you will get as a result. The buyer has to be aware. Than you have to have adequate plans for the site survey, monitoring tools and the intrusion protection system.
Weinschenk: When will four chains be available?
Finneran: A couple of vendors have announced chip sets for four by four. The lag to product availability is about a year probably. If customers are looking for capacity [now], they should be looking at the state of the art today, which is three by three, even though the full transmission rates will go up in the end. We are not yet seeing the last chapter, [although] you can get relatively enormous capacity today with a three by three system, using a 40 MHz system.
Weinschenk: Can you mix and match among the chains that are available, say a transmit side with a capacity of two chains and a receive side with three?
Finneran: Absolutely. The problem is that it will revert to the least common denominator, the highest one that both will support. So it is compatible but it is the limiting factor. From what I found, the vendors are not forthcoming about offering those details unless you ask, so you have to know what you are buying.
Weinschenk: What vendors are key here?
Finneran: The positions are fairly well set. The main part is the infrastructure, whose WLAN systems the organization already is using. Cisco has roughly 65 percent market share. Coming behind them you have Aruba, Motorola, Meru and Trapeze.
Weinschenk: What should people look for when shopping among the vendors?
Finneran: It's highly unlikely that someone would replace their entire infrastructure as part of an upgrade. However, if they were considering a potential "rip and replace," there are a number of factors in N that would be important to consider.
Number one is the overall system architecture, which is either centralized, distributed or a hybrid traffic forwarding approach.
Second, look for enhancements specifically for "n" devices. The critical one we are seeing is traffic scheduling. By giving precedence to high-speed transmitters, we can increase their throughput dramatically with very little negative impact on other users.
The third point is the ability to support "n" access points using traditional power over Ethernet.
But there are not too many green fields any longer. The most likely scenario is people upgrading their existing WLAN with "n" access points.
Weinschenk: Are there significant differences between the products, or are they different versions of essentially the same mousetraps?
Finneran: I don't see anyone making a convincing case that their implementation of "n" is materially better than anyone else's. The possible exception to that would be the new scheduling algorithms, but we have not yet had sufficient real-world experience to determine its magnitude. By letting the fast guys send sooner and more than the slow guys, you can have a real improvement to throughput to fast devices with little performance impact to the slower ones. It's not actually part of "n." It involves differences in how vendor equipment would be implemented. When you have radically different speeds, you can have positive impact by treating different stations differently.
Weinschenk: Sounds like a big deal.
Finneran: The scheduling algorithm only started to appear when higher speed "n" stations started to appear. That was within the last 12 months. Each one is coming up with their own story. The two I heard from initially were Cisco and Aerohive. The impact will be substantial if you believe vendors' white papers. What they are describing clearly is a positive impact. However, white papers always show the best possible outcome. In any case, there will be a notable difference.
Weinschenk: Is there anything else to consider when planning an upgrade?
Finneran: Companies should know how to incorporate it, which means recognizing that you probably need a new site survey to determine which band you are going to implement, whether it is 2.4 or 5 GHz. Our general recommendation is 5 GHz. Then begin planning to migrate existing users to "n."
Weinschenk: How will 802.11n fit into the bigger telecom and IT landscapes?
Finneran: 802.11n was a major technological step forward, a whole new generation ahead in radio technologies. The technologies in 802.11n are very similar to 4G LTE and WiMax approaches. They all use MIMO and OFDM modulation. Absolutely, it's changing the world. I can have a router that has a LTE or WiMax wide-area interface and a 802.11n local interface. Going forward, the two most important interfaces will be LTE and 802.11n.
Weinschenk: So the future is becoming a bit clearer.
Finneran: It's the Darwinian evolution of our technology. The same way "g" devices replaced "b" devices, "n" will replace "a, b" and "g." A device that needs a local- and wide-area interface will have LTE for the wide area and 802.11n for the local interface. That is what we are merging to. Everybody is coming up with the same answer.