Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ Patently Absurd ]   |  You Can't Get There From Here  |   [ Two Steps Forward, Two Steps Back ] >>

Weekly Column

You Can't Get There From Here: Why No Single Networking Technology Is Adequate

Status: [CLOSED]
By Robert X. Cringely

Here is the process I go through when I am trying to learn enough about something in order to write about it with authority. I read all I can, then I talk to people. Talking is important because you can't ask a book or a web site, "Why is that?" or "What about this?" So I find the smartest people I can in a given subject area and ask them questions until they are sick of me. And every time the same thing happens.

The first person I talk with makes so much sense and is so persuasive that I just can't imagine even needing to talk with another person. But I talk to another person anyway. Often, that second person takes a position just the opposite of the first person I talked to, yet I find the second person just as persuasive! How could I have been so easily swayed by the first person? Clearly, this second person has a better view.

I go on to talk with another person and another, each time coming away totally convinced. Then something happens as the gathered information reaches a critical mass and I begin to develop my own view of the topic �- a view I not only believe, but I can do a pretty fair job of defending because of the agonizing way it was developed. The total number of people I talk with depends on how much or little I know about the subject, how complex it is, and how much I plan to write about it. For a typical column, I probably talk with a dozen people. For a book or a documentary, the number is never less than 200 and can go as high as 400. The most important part of this process -- and the reason I am even mentioning it -- is that decision to gather more information, to talk with another person, because sometimes what makes the most sense at first doesn’t turn out to be the correct answer.

This applies very much in the area of wireless networking, where people are grasping at the first viable answer they find, assuming for no reason at all that it is the best. The truth is there is no single best solution.

A few weeks ago, I wrote about the emerging 802.15 networking standard for wirelessly connecting home entertainment devices like televisions, DVDs, and personal computers. After that column appeared I heard from a number of readers who couldn't imagine why I was excited about 802.15 when 802.11g was already available and would work over a much greater distance. Was I just stupid or what?

Then not long ago I had lunch with a guy named Sriram who runs Intel Capital and is funneling hundreds of millions of Intelbucks into various wireless networking schemes that include WiFi hotspots and 802.15 devices. Sriram wasn’t a partisan of 802.11g or 802.15, but of ALL wireless networking schemes at the expense of wired approaches. In Sriram's mind, the future of networking is wireless, and it would be stupid to put money in anything else.

These are both examples of settling for the first viable answer, and deciding that it makes no sense to look further.

Both positions are misguided.

Take 802.11g versus 802.15. Certainly, with 54 megabits-per-second to work with, 802.11g has the bandwidth to carry just about any signal all the way to HDTV's 20 megabits-per-second. Why do we need 802.15 at all? This answer comes down to Quality of Service (QoS), interference and congestion. With 802.11 wireless, all its incarnations offer no true Quality of Service. There is no guarantee that my HDTV signal will get preferential treatment over Grandad's porno downloads.

We can see this most obviously in the 802.11b adapters from several companies that are intended to carry video and audio: What they actually carry are pictures and audio. Companies like Linksys and D-Link are selling these gizmos so you can play MP3s from your hard drive through your stereo system and so you can look at your electronic family photos, but they very specifically make no claims that you can watch streaming video. Why not? Even at 11 megabits-per-second, there ought to be plenty of bandwidth to carry typical Windows Media, Real, or QuickTime video.


The problem here is two-fold. First, there’s that Quality of Service glitch. The second problem is that while your DSL or cable modem may be downloading video that is encoded at 300 kilobits-per-second, what we're talking about is sending that video to your TV, which means it has to first be decoded. If any of these modern encoders can shrink video by 100-to-1, that means the decoding process EXPANDS it by 100-to-1. Your DSL modem is talking to your PC at 300 kilobits-per-second, but your PC is trying to talk to your TV at 30 MEGAbits-per-second. So 802.11b is out of the question for streaming video unless you put the decoder into the TV, which I think will inevitably happen but not this week.

However, 802.11g can theoretically handle that 30 megabit-per-second stream, but even then, it is dodgy. There is still the QoS problem, but even if you were the only person on the network, it still might not work. That's because actual throughput of an individual signal on an 802.11g network is limited to 20 megabits-per-second if there is even a single 802.11b device running on the network, and around 15 megabits-per-second in real world testing.

But wait, there's more! Congestion is also a problem.

Both 802.11b and g are limited to three channels in a contiguous space and 802.11a can handle up to eight channels, but even at the 108 megabits-per-second that some vendors claim to support through full-duplex operation, that means there are at most 324 megabits-per-second available for 802.11g users and 864 megabits-per-second for 802.11a users. Three hundred megabits is a lot of bits until you try sharing them with a few hundred other people in an office, dormitory, or apartment building, and all of them want to watch different video streams. That’s why we won’t see WiFi replacing cable TV anytime soon.

Finally, there is interference. WiFi uses 2.4 Ghz for 802.11b and g and 5.8 Ghz for 802.11a. Both of these frequency bands do not require a license from the Federal Communications Commission as long as your device is compliant with all technical standards AND DOESN'T INTERFERE. This last bit is where some people just don't get it. Say you want to set up a WiFi LAN and your neighbor's 2.4 Ghz baby monitor or cordless phone makes that impossible. According to the rules, they aren't allowed to interfere with you. Say you want to set up a WiFi LAN in a place where there are already lots of other WiFi LANs and it doesn’t look like you can find any channel capacity. According to the rules those other LANs aren’t allowed to interfere with you, either. If they do, you can call the FCC and those other folks may have to pay a fine.

Certainly, they’ll have to make room.

The system demands that everyone play fair, but the system never envisioned technology run amok. Especially in the 2.4 Ghz band, this interference is going to be a growing problem with effective throughput slowing for all users -- yet another reason why we won’t be using WiFi for streaming video.

And that’s why we need 802.15.3 and 3a, which support vastly more users with vastly more bandwidth and vastly less interference. The whole idea of 802.15 is, like Bluetooth, to network one room, not a home or an office, much less a building or a campus. That limits the number of possible devices to a computer, a remote control, and anything in your stereo rack. That means that 802.15.3 supports four channels running at 55 megabits-per-second, each with true Quality of Service. With 802.15.3a, the Ultra Wide Band subvariant, it increases total throughput to 1.3 gigabits-per-second. The area served by a single 802.11g network access point can support more than 70 802.15 networks.

The lure of 802.15 for users is they’ll never have to figure out how to wire their home entertainment system: the network self-discovers and just works. The lure of 802.15 for consumer electronics manufacturers is that we’ll throw away all our old gear and buy new. But this doesn’t mean that wireless networks are appropriate for all devices. How do we get high bandwidth signals to those little networks? That probably requires copper wire or optical fiber. Since nearly all of these devices have to be plugged into power, why not use the power line as a self-discovering network? HomePlug, the new power line networking standard, has been a dud so far, primarily because people are so caught by the wireless dream. But several Japanese companies are planning to add power line networking to their new devices, possibly in combination with 802.15. There are single chipsets being developed, in fact, that support wired, wireless, and power line networking � anything to get us to throw away that old TV.

And there’s the lesson for this Christmas high-tech buying season: There is no one best answer. Our appetite for bandwidth is going to create serious problems for WiFi, but we can’t just trade that for serious problems with some alternate scheme. It will take a combination of networking technologies for us to achieve our ultimate goal of total couch potato-hood.

Comments from the Tribe

Status: [CLOSED] read all comments (0)