What is performance?

Farpoint Group –

You know, we (and especially I) throw the term performance around a lot. In networking, mentioning performance almost always implies throughput, and these two terms are frequently used interchangeably. Performance thus means how fast a given connection is in terms of bits per second.

Unfortunately, that definition is much too limited when we talk about wireless. As regular readers of this column know, the variable nature of the radio channel, and the consequential statistical behavior of wireless links, results in highly-variable throughput, which can range from just fine to nonexistent, and which, of course, can change on a moment-by-moment basis. This is perhaps the greatest frustration of any wireless user - after all, wired performance is highly predictable, especially when wired networks are carefully planned and overprovisioned, as is quite often the case today.

But the wireless environment is a lot tougher, so you never really know what you're going to get until you get it. The good news is that improvements in radio technology continue at what seems to be an accelerating pace, and lower costs allow us to overprovision in a manner analogous to that on wire (see my earlier article and whitepaper on dense deployments for more information). So wireless throughput continues to improve, but variability will almost always be with us as a fact of life.

But there's more to wireless performance than throughput alone. Let's consider some of the other dimensions of wireless performance, all of which influence throughput as well:

  • Range: Those with limited experience in wireless always ask (right after the question about speed, of course) how far a given wireless connection will go, implying that more is better. Rarely, however, is more range better in wireless. The right amount of range is exactly that required to make a reliable connection at the desired level of throughput, but no more. Range beyond this means that we're using radio spectrum we don't need, and thus perhaps interfering with others nearby trying to use the same frequencies. This is especially true in the unlicensed bands used by wireless LANs and PANs. But sometimes more range is needed, especially indoors where the environment can quickly degrade a wireless link. Interestingly, MIMO technology (see my earlier article on the subject) is perhaps more valuable today for the improvements it brings to range than throughput. In yet another earlier article, I described how using MIMO can dramatically improve the quality of a wireless LAN connection when more range is required.
  • Coverage: Especially when we talk about metro-area and wide-area wireless networks, coverage becomes a key element. I know of no one who's never suffered a dropped cell phone call, and the primary reason for this is that there aren't enough cellular base stations to cover a given area. This can sometimes also occur because all of the spectrum provisioned by a given base station is subscribed, leaving no room for new or roaming users, but that also basically means that there aren't enough base stations to provide good coverage. The solution is to this coverage (or availability) problem is to put in more base stations, but this is easier said than done in cellular systems. Base stations are expensive, and involve protracted legal reviews and real-estate expenses. This is one reason why I think the combination of cellular and Wi-Fi is the best solution to the coverage challenge. As long as we're on the subject of Wi-Fi coverage, issues analogous to the problems with cellular networks described above can also arise with Wi-Fi - but these are a lot easier to deal with!
  • Capacity: When we put all of this together, the name of the game is to provision enough overall capacity to handle the traffic demands of the user base in a given location and at a given time. Capacity refers to the ability of a network solution to move the amount and types of data present at any given moment in time. Increasingly, this means being able to handle lots of voice and lots of data simultaneously. This, to me, is the ultimate measure of performance, and it challenges not just wireless components, but the entire network value chain.
  • Latency: Which brings me to latency. We often complain that wireless networks, especially those based on cellular data, are slow (well, I do, anyway). But the over-the-air performance is usually fine; rather, the problem is congestion elsewhere in the network resulting in latency and thus delay. Again, end-to-end network, not just wireless, engineering is the answer. We're always quick to blame radio when things aren't going well, but it's often problems on wire that are the real source of our consternation.

The good news is that overprovisioning, as I noted above, is getting easier as equipment costs continue to decline. The cellular industry is still faced with the problem of siting new base stations, usually due to zoning and NIMBY issues. They'll be with us for some time. But combined cellular/Wi-Fi networks should help.

But the really good news is that new radio technologies beyond MIMO and ultrawideband are beginning to appear; the rate of innovation in wireless appears to be continuing to accelerate. Just two brief examples:

  • There's seven GHz of spectrum up around 60 GHz available for use - and, amazingly, perhaps, it's unlicensed. 60 GHz has a number of challenges, to be sure - it's the band where the oxygen is the air absorbs radio, so range will be limited (but not enormously so), and it's hard to build 60 GHz devices. But both of these issues can be addressed, and we will next year see very, very high-speed (albeit limited-range) wireless devices indeed.
  • There's also some remarkable innovation on modulation techniques going on. For example, a company called XG Technology has developed a scheme that looks a lot at first glance like UWB, but will work over relatively long distances and potentially very high data rates - tens of megabits, depending upon implementation. This could be very competitive with 3G, 4G, and mobile WiMax.

And there are others - I'll have more for you on this subject later. For now, just keep in mind that all dimensions of radio performance can be improved by applying new approaches to radio technology. It just keeps getting better.

What’s wrong? The new clean desk test
Join the discussion
Be the first to comment on this article. Our Commenting Policies