WLAN: 802.11n Impact - Part 1 - Up to 10X Performance

• Report rights infringement
• published: 13 Mar 2012
• views: 8197

Dr. Bharghavan, Founder of Meru Networks, discusses 802.11n in this 8 part series.For more info visit: http://www.MeruNetworks.com One of the most significant developments in enterprise networking in the last decade has been the emergence and recent ratification of 802.11n. 802.11n has the potential to fundamentally change the way the enterprise edge works, not just in specific verticals but in the broad enterprise. The reason is that with 802.11n you now get the capacity, the reliability, and potentially the ability to scale with a very large number of devices with a system that's designed right so that the edge can migrate from primarily wired to primarily wireless. In this pair of videos, what I'm going to do is first talk about the need for 802.11n, provide a very brief overview of what all the different mechanisms are that go into 802.11n, and then the high-level benefits. That will constitute the first video. Really the purpose of the first video is to talk about the evolution of the technology and the standard itself. Describe the benefits, but also highlight the complexity of 802.11n. The second video, which is really a companion video to the first, will tell you how to build a network, because building a network with 802.11n is not as simple as ripping out your legacy abg access points and putting in 11n access points, as you will see. It requires fundamentals trade-offs in terms of technology and in terms of how networks are built. So what we'll do is talk about what these trade-offs are and then go into the design principles that allow one to build an enterprise-class network that has the potential to replace the wired edge with wireless. The first key point is why we need 802.11n at all. If you go back and look at the limitations of 802.11abg, there are fundamentally three. The first one is limited capacity, the second one is channel inefficiency, and the third one is the inability of the standard to handle multipath, in other words, how susceptible it is to RF level loss. The first point is how you deal with an unlimited capacity. If you look at the evolution of the standard again, 802.11b gave you about 6 Mbps of end- to-end throughput, as opposed to 11Mbps peak data rate. With a and g, you got about 20Mbps of throughput, give or take, on a peak data rate of 54Mbps. But with 802.11n, we now have the ability to get about 200Mbps of end-to-end TCP throughput. This is about twice as fast as what you can get with Fast Ethernet, and therefore fundamentally changes the equation. Going to the second key point of the limitation of 802.11abg, it's not just that capacity was limited. It's that the way the channel was utilized itself was inherently inefficient. The amount of time that was spent in contention and in header overhead was significant. Again, remember from our past videos that 802.11abg is fundamentally a hub standard. In a hub, as the number of devices increase, a lot of the airtime is spent wasted in contention. In abg in particular, you would content for every frame. So the contention overhead itself started to take over as the number of devices started to increase. On top of that, the headers are sent at very low data rates. Therefore, if you look at the total channel efficiency, you will see that you can reach a peak of about 50 to 60 percent. That in part explains why a standard that at the MAC level gives you 54Mbps of data rate is only able to deliver an end-to-end throughput of about 20Mbps of peak TCP throughput. The third key aspect with 802.11ag was really the way the physical layer and the RF layer were structured. The technology was such that it was highly susceptible to multipath. Now, if you're looking at any kind of indoor environment, notice that signals bounce off walls. Therefore, any transmission can be heard by a receiver not just from one path but from multiple paths, depending upon the number of walls they bounce off of. So in an indoor environment, signal might reflect, it might refract, and it might dissipate. As a result, the receiver actually gets signals that are somewhat corrupted, and it was up to the receiver in order to figure out or essentially decode legitimate signal from corrupted signal. Because of the limitations of the technology, this concept of signal fidelity or the level of signal fidelity that abg technology provided was significantly lower than 11n. What we'll do is now talk about how 11n improves the signal fidelity. As a result, with an abg network, the further you go, the more corrupt the signal gets, the lower the signal fidelity, and hence, lower the data rate. At this point, I think we've talked about the three key takeaways in terms of limitation of abg, which brought us to 11n, which is: (1) limited capacity; (2) inefficient use of that capacity; and (3) multipath, which effectively reduces the capacity or signal quality, particularly with distance and indoor environments. To learn more visit: http://www.MeruNetworks.com