WISP How to: Detect and Avoid Self-Inflicted Interference

WISP How to: Detect and Avoid Self-Inflicted Interference

Photo of author
Written By Eric Sandler

By Marlon Schafer

July 1, 2008

We’ve all seen it. Well, we can’t actually *see* it. We certainly do see the results of it, though. It will wreck your network’s throughput, sometimes even causing outages. It’s the scourge of the unlicensed world. That’s right, I’m talking about Interference. Capital I. Catastrophic destruction of your network and your reputation.

What’s the first thing we ever do when we finally figure out this is what’s happening? We blame the dirtbag competitor that’s also on the tower or somewhere nearby. After all, things were running just grand until Mr. Dirtbag came along.

For years I’ve warned people about the worst kind of interference though. The self inflicted kind. It’s the easiest to test for, but often the hardest to diagnose. We tend to look everywhere else first. After all, nothing has changed on my network. Right?

Well, you’ve added more customers right? And those customers are using more and more access all of the time. (Remember how people started using more and more time on the old dial-in systems??? Same thing, different technology.)

Here is a look at one of my sites back in September of 2003.

Five APs detectable, and two of them are mine (accima).

Here’s the same site in October of that year:

As you can see, there were about 18 APs in a 360° sweep of the area. Three of those are other systems that I had at that same site. Over the years, as I’ve built up 30 transmit sites and over 6000 square miles of coverage I’ve had to narrow my AP’s field of view to the point that some only have 30° of coverage!

And in May of 2007:

Now there are over 50 APs and we’re now only looking at a 45° sector pointed to the side of town! Four of these are mine.

Here’s where that gets interesting. Look at the signal levels in the right hand column. Assuming I’ve done a good job and my customers are hitting the AP in the -65 to -75 range I’m still mostly OK with the 15dB of carrier to interference ratio (called C to I). It’s close, but most of the interference will be too low to matter.

Except my own. Anything that says “accima” is mine. Now we suddenly have -65, -71, -65 and -84 signal levels!!!! At -84 on the main signal we can probably do a good job of isolating one from the other with channel spacing, even if we don’t use only channels 1, 6, and 11. But with -65 we will have overlap. For a very good article on some of these issues check out this article by Carl Andren from Harris Semiconductor.

Why, really, does that matter so much? Well, you have to understand how Wi-Fi systems work (and even how many non Wi-Fi systems work). They have a mechanism called CSMAK. This basically says that the airways must first be clear, then I can transmit my data. If there is no clear air, I will wait a random amount of time then try again. Eventually the competing system will be out of the way and I’ll get my data through.

However, when we have busier and busier systems (both our own and others in the area) there is less and less clear air. Back when I had 5 or 10 people on an AP I could just point a new AP in a new direction, give it a new name, and then service customers from it, even when using the same channel. That just don’t work no more! There ain’t nough clear air.

But, you say, there are 11 channels! Sorry, but the radios don’t drop off at the edge of a channel. The power levels drift downward as you move farther from the frequency center. Here’s an example of that overlap.

Notice how much overlap there is. And, realistically, in most areas these days we don’t have use of channel 6 or those surrounding it. So we’re really down to 2 channels that are non-overlapping at all.

We also have to take into account how much more sensitive the hardware is these days. I have a site with a 120° sector (13dB) that picks up a tower that’s only putting out 35dB (just under the 4 watt legal power level for an AP) from over 30 miles away!!!! Self inflicted interference isn’t just limited to devices on the same tower anymore.

So, how do you test for this? Well, first off, try changing your channels around a bit. Put all of your ap’s in an area on channel 11 and see if moving one at a time to channel 1 makes those using channel 1 work any better than before. (We run tests to http://blt.speakeasy.net.) Or, just try moving one at a time to different channels. We use any channels, anywhere. Just don’t forget about interference between your CPE units and the home routers your customers use. You can’t really use channels 5, 6, and 7 in most cases because of the customer’s routers. Usually 4 and 8 stink too. They might work fine for some of your stuff or maybe backhauls, but usually I just stay away from them.

The other good test is to unplug all but one AP and see what the before and after tests show. I know that option pretty well sucks. But it is something that you have control over so it’s a good place to start when you’ve tried everything else you can think of.

Unfortunately, these days, the airways are incredibly crowded. And in our rush to pull out omni antennas and sectorize, we end up needing to use more spectrum than is available today. Yeah, you will have to sectorize to help minimize the interference that you’ll see. Just don’t forget about the closest source of it.

Is it still worth sectorizing busy sites? You bet. I typically see my speeds at least double after I sectorize. Sometimes I see this just when I put newer radios up at the antenna and ditch the amps. I have customers getting over 2 Mbps down and up at 12+ miles. The base station is a 250mw radio and an 8dB omni. I’ve got one link that does even better than that with a 15 or 17 dB radio into a 13dB sector—18 miles! We can go much farther with much less power than ever before. It just takes more creativity and talent than it used to.

Stay safe out there!

Marlon K. Schafer

Leave a Comment